Colon and Rectal

SAN FRANCISCO – Opioids may not always be necessary following elective colorectal surgery. That’s the message coming from a retrospective study of medical records at the University of Massachusetts Medical Center, Worcester, which found that over half of patients never even filled out their prescription after colorectal surgery.

“We found that over half of the patients took no opioid pills after discharge, and 60% of the prescribed pills were left over,” said David Meyer, MD, during a presentation of the study at the annual clinical congress of the American College of Surgeons. Dr. Meyer is a surgical resident at the University of Massachusetts.

The team also used the results of their analysis to develop a guideline for the amount of opioid to prescribe following major colorectal surgery, with specific amounts of pills recommended based on the amount of opioid use during the last 24 hours of hospitalization.

“It shows a real interest in tailoring our postoperative care in pain management. They’re trying to find a way to hit a sweet spot to get patients the right amount of pain control,” said Jonathan Mitchem, MD, in an interview. Dr. Mitchem is an assistant professor at University of Missouri–Columbia, and comoderated the session where the research was presented.

The researchers performed a retrospective analysis of major elective colorectal procedures at their institution, including colectomy, rectal resection, and ostomy reversal. The analysis included 100 patients (55 female), with a mean age of 59 years. A total of 71% were opioid naive, meaning there was no evidence of an opioid prescription in the year prior to surgery. A total of 74% underwent a laparoscopic procedure, and 75% had a partial colectomy. The postoperative stay averaged 4.5 days.

The researchers converted in-hospital opioid use categories (IOUC) to equianalgesic 5-mg oxycodone pills (EOPs). In the last 24 hours before release, 53% of patients had no opioids at all (no IOUC, 0 EOPs), 25% received low amounts of opioids (low IOUC, 0.1-3.0 EOPs), and 22% high amounts (high IOUC, more than 3.1 EOPs). Overall, prescribed EOP was 17.5, and just 38% was consumed. These numbers were lowest in the no-IOUC group (15.7, 16%), followed by the low-IOUC group (16.0, 32%), and the high group (23.7, 79%; P less than .01).

The researchers then looked at the 85th percentile of EOPs for each group, and used that to develop a guideline for opioid prescription. For the no-IOUC group, they recommend 3 EOPs, for the low-IOUC group they recommend 12 EOPs, and for the high-IOUC group they recommend 30 EOPs.

The researchers examined various factors that might have influenced opioid use, correcting for whether the patient was opioid naive, case type, postoperative length of stay, and new ostomy creation. The only factor with a significant association for excessive opioid use was inflammatory bowel disease, which was linked to a nearly 900% increased risk of using more than the guideline amounts (adjusted odds ratio, 8.3; P less than .01; area under the curve, 0.85).

The study is limited by the fact that it was conducted at a single center, and that patient opioid use was self-reported. The guidelines need to be validated prospectively.

No funding information was disclosed. Dr. Meyer and Dr. Mitchem had no relevant financial disclosures.

SOURCE: Meyer D et al. Clinical Congress 2019, Abstract.

SAN FRANCISCO – Opioids may not always be necessary following elective colorectal surgery. That’s the message coming from a retrospective study of medical records at the University of Massachusetts Medical Center, Worcester, which found that over half of patients never even filled out their prescription after colorectal surgery.

“We found that over half of the patients took no opioid pills after discharge, and 60% of the prescribed pills were left over,” said David Meyer, MD, during a presentation of the study at the annual clinical congress of the American College of Surgeons. Dr. Meyer is a surgical resident at the University of Massachusetts.

The team also used the results of their analysis to develop a guideline for the amount of opioid to prescribe following major colorectal surgery, with specific amounts of pills recommended based on the amount of opioid use during the last 24 hours of hospitalization.

“It shows a real interest in tailoring our postoperative care in pain management. They’re trying to find a way to hit a sweet spot to get patients the right amount of pain control,” said Jonathan Mitchem, MD, in an interview. Dr. Mitchem is an assistant professor at University of Missouri–Columbia, and comoderated the session where the research was presented.

The researchers performed a retrospective analysis of major elective colorectal procedures at their institution, including colectomy, rectal resection, and ostomy reversal. The analysis included 100 patients (55 female), with a mean age of 59 years. A total of 71% were opioid naive, meaning there was no evidence of an opioid prescription in the year prior to surgery. A total of 74% underwent a laparoscopic procedure, and 75% had a partial colectomy. The postoperative stay averaged 4.5 days.

The researchers converted in-hospital opioid use categories (IOUC) to equianalgesic 5-mg oxycodone pills (EOPs). In the last 24 hours before release, 53% of patients had no opioids at all (no IOUC, 0 EOPs), 25% received low amounts of opioids (low IOUC, 0.1-3.0 EOPs), and 22% high amounts (high IOUC, more than 3.1 EOPs). Overall, prescribed EOP was 17.5, and just 38% was consumed. These numbers were lowest in the no-IOUC group (15.7, 16%), followed by the low-IOUC group (16.0, 32%), and the high group (23.7, 79%; P less than .01).

The researchers then looked at the 85th percentile of EOPs for each group, and used that to develop a guideline for opioid prescription. For the no-IOUC group, they recommend 3 EOPs, for the low-IOUC group they recommend 12 EOPs, and for the high-IOUC group they recommend 30 EOPs.

The researchers examined various factors that might have influenced opioid use, correcting for whether the patient was opioid naive, case type, postoperative length of stay, and new ostomy creation. The only factor with a significant association for excessive opioid use was inflammatory bowel disease, which was linked to a nearly 900% increased risk of using more than the guideline amounts (adjusted odds ratio, 8.3; P less than .01; area under the curve, 0.85).

The study is limited by the fact that it was conducted at a single center, and that patient opioid use was self-reported. The guidelines need to be validated prospectively.

No funding information was disclosed. Dr. Meyer and Dr. Mitchem had no relevant financial disclosures.

SOURCE: Meyer D et al. Clinical Congress 2019, Abstract.

SAN FRANCISCO – Opioids may not always be necessary following elective colorectal surgery. That’s the message coming from a retrospective study of medical records at the University of Massachusetts Medical Center, Worcester, which found that over half of patients never even filled out their prescription after colorectal surgery.

“We found that over half of the patients took no opioid pills after discharge, and 60% of the prescribed pills were left over,” said David Meyer, MD, during a presentation of the study at the annual clinical congress of the American College of Surgeons. Dr. Meyer is a surgical resident at the University of Massachusetts.

The team also used the results of their analysis to develop a guideline for the amount of opioid to prescribe following major colorectal surgery, with specific amounts of pills recommended based on the amount of opioid use during the last 24 hours of hospitalization.

“It shows a real interest in tailoring our postoperative care in pain management. They’re trying to find a way to hit a sweet spot to get patients the right amount of pain control,” said Jonathan Mitchem, MD, in an interview. Dr. Mitchem is an assistant professor at University of Missouri–Columbia, and comoderated the session where the research was presented.

The researchers performed a retrospective analysis of major elective colorectal procedures at their institution, including colectomy, rectal resection, and ostomy reversal. The analysis included 100 patients (55 female), with a mean age of 59 years. A total of 71% were opioid naive, meaning there was no evidence of an opioid prescription in the year prior to surgery. A total of 74% underwent a laparoscopic procedure, and 75% had a partial colectomy. The postoperative stay averaged 4.5 days.

The researchers converted in-hospital opioid use categories (IOUC) to equianalgesic 5-mg oxycodone pills (EOPs). In the last 24 hours before release, 53% of patients had no opioids at all (no IOUC, 0 EOPs), 25% received low amounts of opioids (low IOUC, 0.1-3.0 EOPs), and 22% high amounts (high IOUC, more than 3.1 EOPs). Overall, prescribed EOP was 17.5, and just 38% was consumed. These numbers were lowest in the no-IOUC group (15.7, 16%), followed by the low-IOUC group (16.0, 32%), and the high group (23.7, 79%; P less than .01).

The researchers then looked at the 85th percentile of EOPs for each group, and used that to develop a guideline for opioid prescription. For the no-IOUC group, they recommend 3 EOPs, for the low-IOUC group they recommend 12 EOPs, and for the high-IOUC group they recommend 30 EOPs.

The researchers examined various factors that might have influenced opioid use, correcting for whether the patient was opioid naive, case type, postoperative length of stay, and new ostomy creation. The only factor with a significant association for excessive opioid use was inflammatory bowel disease, which was linked to a nearly 900% increased risk of using more than the guideline amounts (adjusted odds ratio, 8.3; P less than .01; area under the curve, 0.85).

The study is limited by the fact that it was conducted at a single center, and that patient opioid use was self-reported. The guidelines need to be validated prospectively.

No funding information was disclosed. Dr. Meyer and Dr. Mitchem had no relevant financial disclosures.

SOURCE: Meyer D et al. Clinical Congress 2019, Abstract.

SAN FRANCISCO – When it comes to KRAS mutational status, liver metastases originating from left-sided colon tumors have different clinical and survival characteristics from those originating from primary rectal tumors. There was no significant difference in survival between bearers of mutated versus wild-type (WT) KRAS in rectal tumor cases, while there was a significant difference in survival from left-sided colon cases, according to a first-time analysis of the effect of KRAS status in this specific population.

The work was presented at the annual clinical congress of the American College of Surgeons by Neda Amini, MD. “The liver metastasis originating from a rectal tumor might have a different biology than from a primary colon tumor, and we should have stratification according to the primary tumor location in clinical trials testing chemotherapy and targeted agents,” said Dr. Amini, who is a surgical resident at Sinai Hospital of Baltimore, during her presentation of the research.

“I thought that was interesting because most of the studies that have been done look at KRAS mutations in colorectal cancers, and [colon and rectal cancers] are two completely different entities,” said session comoderator Valentine Nfonsam, MD, associate professor of surgery at the University of Arizona, Tucson, in an interview. The findings could also impact clinical practice. “If a patient has rectal cancer, if they have a KRAS mutation, whether you treat them with cetuximab or not, the overall survival doesn’t really change. Whereas for colon cancer patients you really want to make that distinction. You want to truly personalize their therapy, because of the difference in survival in a patient with the KRAS mutation in colon cancer” Dr. Nfonsam said.

“It gets to the heart that there might be different biology between colon cancers and rectal cancers. It’s important to understand the differences in the basic biology, which affects the treatment and the surgery,” said the other comoderator, Jonathan Mitchem, MD, in an interview. Dr. Mitchem is an assistant professor at the University of Missouri–Columbia.

KRAS is common in colorectal cancer, occurring in 30% of cases, and multiple trials have shown it is associated with nonresponse to the epidermal growth factor receptor inhibitors cetuximab or panitumumab. All colorectal cancer patients with liver metastases should be screened for KRAS mutations, according to National Comprehensive Cancer Network guidelines.

The researchers conducted a retrospective analysis of 1,304 patients who underwent curative-intent surgery for colorectal liver metastases at nine institutions between 2000 and 2016. The KRAS mutation rate was similar in the primary colon and rectal tumors (34.2% vs. 30.9%; P = .24). The frequency was highest in right-sided colon tumors (39.4%). There was a statistically significant difference in the frequency of KRAS mutation between primary rectal tumors (30.9%), and left-sided colon tumors (21.1%; P = .001).

There were several differences in clinical characteristics between left-sided colon cancers and rectal cancers. Rectal cancer patients were more likely to be male (73.4% vs. 62.4%; P = .001); more likely to be stage T1-T2 (16.6% vs. 10.6%; P = .012); less likely to have serum carcinoembryonic antigen greater than 100 ng/mL (8.4% vs. 14.1%; P = .018); and less likely to have a liver metastasis under 3 cm (36.1% vs. 49.3%).

There were significant differences between KRAS mutant and KRAS wild-type patients with a colon primary tumor, including greater likelihood of lymph node metastasis in WT (65.2% vs. 55.37%; P = .004), greater likelihood of liver metastasis greater than 3 cm in WT (48.8% versus 39.3%; P = .01), greater likelihood of extrahepatic disease in mutant KRAS (16.3% vs. 10.4%; P = .01), greater likelihood of prehepatic resection chemotherapy in WT (65.5% vs. 56.0%; P = .005), greater likelihood of posthepatic resection chemotherapy in mutant KRAS (64.5% vs. 55.5%; P = .01), and greater likelihood of receiving anti–epidermal growth factor therapy in WT (5.7% vs. 0.3%; P less than .001). The only difference seen in patients with rectal primary tumors was the odds of receiving post-hepatic surgery chemotherapy, which was higher among patients with mutated KRAS (70.8% vs. 59.0%; P = .03).

After a median follow-up of 26.4 months, the 1-, 3-, and 5-year overall survival rates were 88.9%, 62.5%, and 44.5%. Among patients with primary colon cancer, there was a statistically significant lower survival curve in patients with a KRAS mutation overall and in those with left-sided colon tumors (log rank P less than .001 for both), but there was no significant survival difference between mutation bearers and wild-type patients with a primary rectal tumor (log rank P = .53). A multivariate analysis showed an 82% risk of death from KRAS mutation in primary colon cancer (hazard ratio, 1.82; P less than .001), but a univariate analysis showed no significant mortality association in rectal primary tumors (hazard ratio, 1.13; P = .46).

The funding source was not disclosed. The authors had no relevant financial disclosures.

SOURCE: Amini N et al. J Am Coll Surg. 2019 Oct;229(4):Suppl 1, S69-70.

SAN FRANCISCO – When it comes to KRAS mutational status, liver metastases originating from left-sided colon tumors have different clinical and survival characteristics from those originating from primary rectal tumors. There was no significant difference in survival between bearers of mutated versus wild-type (WT) KRAS in rectal tumor cases, while there was a significant difference in survival from left-sided colon cases, according to a first-time analysis of the effect of KRAS status in this specific population.

The work was presented at the annual clinical congress of the American College of Surgeons by Neda Amini, MD. “The liver metastasis originating from a rectal tumor might have a different biology than from a primary colon tumor, and we should have stratification according to the primary tumor location in clinical trials testing chemotherapy and targeted agents,” said Dr. Amini, who is a surgical resident at Sinai Hospital of Baltimore, during her presentation of the research.

“I thought that was interesting because most of the studies that have been done look at KRAS mutations in colorectal cancers, and [colon and rectal cancers] are two completely different entities,” said session comoderator Valentine Nfonsam, MD, associate professor of surgery at the University of Arizona, Tucson, in an interview. The findings could also impact clinical practice. “If a patient has rectal cancer, if they have a KRAS mutation, whether you treat them with cetuximab or not, the overall survival doesn’t really change. Whereas for colon cancer patients you really want to make that distinction. You want to truly personalize their therapy, because of the difference in survival in a patient with the KRAS mutation in colon cancer” Dr. Nfonsam said.

“It gets to the heart that there might be different biology between colon cancers and rectal cancers. It’s important to understand the differences in the basic biology, which affects the treatment and the surgery,” said the other comoderator, Jonathan Mitchem, MD, in an interview. Dr. Mitchem is an assistant professor at the University of Missouri–Columbia.

KRAS is common in colorectal cancer, occurring in 30% of cases, and multiple trials have shown it is associated with nonresponse to the epidermal growth factor receptor inhibitors cetuximab or panitumumab. All colorectal cancer patients with liver metastases should be screened for KRAS mutations, according to National Comprehensive Cancer Network guidelines.

The researchers conducted a retrospective analysis of 1,304 patients who underwent curative-intent surgery for colorectal liver metastases at nine institutions between 2000 and 2016. The KRAS mutation rate was similar in the primary colon and rectal tumors (34.2% vs. 30.9%; P = .24). The frequency was highest in right-sided colon tumors (39.4%). There was a statistically significant difference in the frequency of KRAS mutation between primary rectal tumors (30.9%), and left-sided colon tumors (21.1%; P = .001).

There were several differences in clinical characteristics between left-sided colon cancers and rectal cancers. Rectal cancer patients were more likely to be male (73.4% vs. 62.4%; P = .001); more likely to be stage T1-T2 (16.6% vs. 10.6%; P = .012); less likely to have serum carcinoembryonic antigen greater than 100 ng/mL (8.4% vs. 14.1%; P = .018); and less likely to have a liver metastasis under 3 cm (36.1% vs. 49.3%).

There were significant differences between KRAS mutant and KRAS wild-type patients with a colon primary tumor, including greater likelihood of lymph node metastasis in WT (65.2% vs. 55.37%; P = .004), greater likelihood of liver metastasis greater than 3 cm in WT (48.8% versus 39.3%; P = .01), greater likelihood of extrahepatic disease in mutant KRAS (16.3% vs. 10.4%; P = .01), greater likelihood of prehepatic resection chemotherapy in WT (65.5% vs. 56.0%; P = .005), greater likelihood of posthepatic resection chemotherapy in mutant KRAS (64.5% vs. 55.5%; P = .01), and greater likelihood of receiving anti–epidermal growth factor therapy in WT (5.7% vs. 0.3%; P less than .001). The only difference seen in patients with rectal primary tumors was the odds of receiving post-hepatic surgery chemotherapy, which was higher among patients with mutated KRAS (70.8% vs. 59.0%; P = .03).

After a median follow-up of 26.4 months, the 1-, 3-, and 5-year overall survival rates were 88.9%, 62.5%, and 44.5%. Among patients with primary colon cancer, there was a statistically significant lower survival curve in patients with a KRAS mutation overall and in those with left-sided colon tumors (log rank P less than .001 for both), but there was no significant survival difference between mutation bearers and wild-type patients with a primary rectal tumor (log rank P = .53). A multivariate analysis showed an 82% risk of death from KRAS mutation in primary colon cancer (hazard ratio, 1.82; P less than .001), but a univariate analysis showed no significant mortality association in rectal primary tumors (hazard ratio, 1.13; P = .46).

The funding source was not disclosed. The authors had no relevant financial disclosures.

SOURCE: Amini N et al. J Am Coll Surg. 2019 Oct;229(4):Suppl 1, S69-70.

SAN FRANCISCO – When it comes to KRAS mutational status, liver metastases originating from left-sided colon tumors have different clinical and survival characteristics from those originating from primary rectal tumors. There was no significant difference in survival between bearers of mutated versus wild-type (WT) KRAS in rectal tumor cases, while there was a significant difference in survival from left-sided colon cases, according to a first-time analysis of the effect of KRAS status in this specific population.

The work was presented at the annual clinical congress of the American College of Surgeons by Neda Amini, MD. “The liver metastasis originating from a rectal tumor might have a different biology than from a primary colon tumor, and we should have stratification according to the primary tumor location in clinical trials testing chemotherapy and targeted agents,” said Dr. Amini, who is a surgical resident at Sinai Hospital of Baltimore, during her presentation of the research.

“I thought that was interesting because most of the studies that have been done look at KRAS mutations in colorectal cancers, and [colon and rectal cancers] are two completely different entities,” said session comoderator Valentine Nfonsam, MD, associate professor of surgery at the University of Arizona, Tucson, in an interview. The findings could also impact clinical practice. “If a patient has rectal cancer, if they have a KRAS mutation, whether you treat them with cetuximab or not, the overall survival doesn’t really change. Whereas for colon cancer patients you really want to make that distinction. You want to truly personalize their therapy, because of the difference in survival in a patient with the KRAS mutation in colon cancer” Dr. Nfonsam said.

“It gets to the heart that there might be different biology between colon cancers and rectal cancers. It’s important to understand the differences in the basic biology, which affects the treatment and the surgery,” said the other comoderator, Jonathan Mitchem, MD, in an interview. Dr. Mitchem is an assistant professor at the University of Missouri–Columbia.

KRAS is common in colorectal cancer, occurring in 30% of cases, and multiple trials have shown it is associated with nonresponse to the epidermal growth factor receptor inhibitors cetuximab or panitumumab. All colorectal cancer patients with liver metastases should be screened for KRAS mutations, according to National Comprehensive Cancer Network guidelines.

The researchers conducted a retrospective analysis of 1,304 patients who underwent curative-intent surgery for colorectal liver metastases at nine institutions between 2000 and 2016. The KRAS mutation rate was similar in the primary colon and rectal tumors (34.2% vs. 30.9%; P = .24). The frequency was highest in right-sided colon tumors (39.4%). There was a statistically significant difference in the frequency of KRAS mutation between primary rectal tumors (30.9%), and left-sided colon tumors (21.1%; P = .001).

There were several differences in clinical characteristics between left-sided colon cancers and rectal cancers. Rectal cancer patients were more likely to be male (73.4% vs. 62.4%; P = .001); more likely to be stage T1-T2 (16.6% vs. 10.6%; P = .012); less likely to have serum carcinoembryonic antigen greater than 100 ng/mL (8.4% vs. 14.1%; P = .018); and less likely to have a liver metastasis under 3 cm (36.1% vs. 49.3%).

There were significant differences between KRAS mutant and KRAS wild-type patients with a colon primary tumor, including greater likelihood of lymph node metastasis in WT (65.2% vs. 55.37%; P = .004), greater likelihood of liver metastasis greater than 3 cm in WT (48.8% versus 39.3%; P = .01), greater likelihood of extrahepatic disease in mutant KRAS (16.3% vs. 10.4%; P = .01), greater likelihood of prehepatic resection chemotherapy in WT (65.5% vs. 56.0%; P = .005), greater likelihood of posthepatic resection chemotherapy in mutant KRAS (64.5% vs. 55.5%; P = .01), and greater likelihood of receiving anti–epidermal growth factor therapy in WT (5.7% vs. 0.3%; P less than .001). The only difference seen in patients with rectal primary tumors was the odds of receiving post-hepatic surgery chemotherapy, which was higher among patients with mutated KRAS (70.8% vs. 59.0%; P = .03).

After a median follow-up of 26.4 months, the 1-, 3-, and 5-year overall survival rates were 88.9%, 62.5%, and 44.5%. Among patients with primary colon cancer, there was a statistically significant lower survival curve in patients with a KRAS mutation overall and in those with left-sided colon tumors (log rank P less than .001 for both), but there was no significant survival difference between mutation bearers and wild-type patients with a primary rectal tumor (log rank P = .53). A multivariate analysis showed an 82% risk of death from KRAS mutation in primary colon cancer (hazard ratio, 1.82; P less than .001), but a univariate analysis showed no significant mortality association in rectal primary tumors (hazard ratio, 1.13; P = .46).

The funding source was not disclosed. The authors had no relevant financial disclosures.

SOURCE: Amini N et al. J Am Coll Surg. 2019 Oct;229(4):Suppl 1, S69-70.

Patients 50-79 years old with a demonstrably low risk of developing the disease within 15 years probably don’t need to be screened for colorectal cancer. But if their risk of disease is at least 3% over 15 years, patients should be screened, Lise M. Helsingen, MD, and colleagues wrote in BMJ (2019;367:l5515 doi: 10.1136/bmj.l5515).

For these patients, “We suggest screening with one of the four screening options: fecal immunochemical test (FIT) every year, FIT every 2 years, a single sigmoidoscopy, or a single colonoscopy,” wrote Dr. Helsingen of the University of Oslo, and her team.

She chaired a 22-member international panel that developed a collaborative effort from the MAGIC research and innovation program as a part of the BMJ Rapid Recommendations project. The team reviewed 12 research papers comprising almost 1.4 million patients from Denmark, Italy, the Netherlands, Norway, Poland, Spain, Sweden, the United Kingdom, and the United States. Follow-up ranged from 0 to 19.5 years for colorectal cancer incidence and up to 30 years for mortality.

Because of the dearth of relevant data in some studies, however, the projected outcomes had to be simulated, with benefits and harms calculations based on 100% screening adherence. However, the team noted, it’s impossible to achieve complete adherence. Most studies of colorectal screening don’t exceed a 50% adherence level.

“All the modeling data are of low certainty. It is a useful indication, but there is a high chance that new evidence will show a smaller or larger benefit, which in turn may alter these recommendations.”

Compared with no screening, all four screening models reduced the risk of colorectal cancer mortality to a similar level.

• FIT every year, 59%.
• FIT every 2 years, 50%.
• Single sigmoidoscopy, 52%.
• Single colonoscopy, 67%.

Screening had less of an impact on reducing the incidence of colorectal cancer:

• FIT every 2 years, 0.05%.
• FIT every year, 0.15%.
• Single sigmoidoscopy, 27%.
• Single colonoscopy, 34%.

The panel also assessed potential harms. Among almost 1 million patients, the colonoscopy-related mortality rate was 0.03 per 1,000 procedures. The perforation rate was 0.8 per 1,000 colonoscopies after a positive fecal test, and 1.4 per 1,000 screened with sigmoidoscopy. The bleeding rate was 1.9 per 1,000 colonoscopies performed after a positive fecal test, and 3-4 per 1,000 screened with sigmoidoscopy.

Successful implementation of these recommendations hinges on accurate risk assessment, however. The team recommended the QCancer platform as “one of the best performing models for both men and women.”

The calculator includes age, sex, ethnicity, smoking status, alcohol use, family history of gastrointestinal cancer, personal history of other cancers, diabetes, ulcerative colitis, colonic polyps, and body mass index.

“We suggest this model because it is available as an online calculator; includes only risk factors available in routine health care; has been validated in a population separate from the derivation population; has reasonable discriminatory ability; and has a good fit between predicted and observed outcomes. In addition, it is the only online risk calculator we know of that predicts risk over a 15-year time horizon.”

The team stressed that their recommendations can’t be applied to all patients. Because evidence for both screening recommendations was weak – largely because of the dearth of supporting data – patients and physicians should cocreate a personalized screening plan.

“Several factors influence individuals’ decisions whether to be screened, even when they are presented with the same information,” the authors said. These include variation in an individual’s values and preferences, a close balance of benefits versus harms and burdens, and personal preference.

“Some individuals may value a minimally invasive test such as FIT, and the possibility of invasive screening with colonoscopy might put them off screening altogether. Those who most value preventing colorectal cancer or avoiding repeated testing are likely to choose sigmoidoscopy or colonoscopy.”

The authors had no financial conflicts of interest.

SOURCE: BMJ 2019;367:l5515. doi: 10.1136/bmj.l5515.

Patients 50-79 years old with a demonstrably low risk of developing the disease within 15 years probably don’t need to be screened for colorectal cancer. But if their risk of disease is at least 3% over 15 years, patients should be screened, Lise M. Helsingen, MD, and colleagues wrote in BMJ (2019;367:l5515 doi: 10.1136/bmj.l5515).

For these patients, “We suggest screening with one of the four screening options: fecal immunochemical test (FIT) every year, FIT every 2 years, a single sigmoidoscopy, or a single colonoscopy,” wrote Dr. Helsingen of the University of Oslo, and her team.

She chaired a 22-member international panel that developed a collaborative effort from the MAGIC research and innovation program as a part of the BMJ Rapid Recommendations project. The team reviewed 12 research papers comprising almost 1.4 million patients from Denmark, Italy, the Netherlands, Norway, Poland, Spain, Sweden, the United Kingdom, and the United States. Follow-up ranged from 0 to 19.5 years for colorectal cancer incidence and up to 30 years for mortality.

Because of the dearth of relevant data in some studies, however, the projected outcomes had to be simulated, with benefits and harms calculations based on 100% screening adherence. However, the team noted, it’s impossible to achieve complete adherence. Most studies of colorectal screening don’t exceed a 50% adherence level.

“All the modeling data are of low certainty. It is a useful indication, but there is a high chance that new evidence will show a smaller or larger benefit, which in turn may alter these recommendations.”

Compared with no screening, all four screening models reduced the risk of colorectal cancer mortality to a similar level.

• FIT every year, 59%.
• FIT every 2 years, 50%.
• Single sigmoidoscopy, 52%.
• Single colonoscopy, 67%.

Screening had less of an impact on reducing the incidence of colorectal cancer:

• FIT every 2 years, 0.05%.
• FIT every year, 0.15%.
• Single sigmoidoscopy, 27%.
• Single colonoscopy, 34%.

The panel also assessed potential harms. Among almost 1 million patients, the colonoscopy-related mortality rate was 0.03 per 1,000 procedures. The perforation rate was 0.8 per 1,000 colonoscopies after a positive fecal test, and 1.4 per 1,000 screened with sigmoidoscopy. The bleeding rate was 1.9 per 1,000 colonoscopies performed after a positive fecal test, and 3-4 per 1,000 screened with sigmoidoscopy.

Successful implementation of these recommendations hinges on accurate risk assessment, however. The team recommended the QCancer platform as “one of the best performing models for both men and women.”

The calculator includes age, sex, ethnicity, smoking status, alcohol use, family history of gastrointestinal cancer, personal history of other cancers, diabetes, ulcerative colitis, colonic polyps, and body mass index.

“We suggest this model because it is available as an online calculator; includes only risk factors available in routine health care; has been validated in a population separate from the derivation population; has reasonable discriminatory ability; and has a good fit between predicted and observed outcomes. In addition, it is the only online risk calculator we know of that predicts risk over a 15-year time horizon.”

The team stressed that their recommendations can’t be applied to all patients. Because evidence for both screening recommendations was weak – largely because of the dearth of supporting data – patients and physicians should cocreate a personalized screening plan.

“Several factors influence individuals’ decisions whether to be screened, even when they are presented with the same information,” the authors said. These include variation in an individual’s values and preferences, a close balance of benefits versus harms and burdens, and personal preference.

“Some individuals may value a minimally invasive test such as FIT, and the possibility of invasive screening with colonoscopy might put them off screening altogether. Those who most value preventing colorectal cancer or avoiding repeated testing are likely to choose sigmoidoscopy or colonoscopy.”

The authors had no financial conflicts of interest.

SOURCE: BMJ 2019;367:l5515. doi: 10.1136/bmj.l5515.

Patients 50-79 years old with a demonstrably low risk of developing the disease within 15 years probably don’t need to be screened for colorectal cancer. But if their risk of disease is at least 3% over 15 years, patients should be screened, Lise M. Helsingen, MD, and colleagues wrote in BMJ (2019;367:l5515 doi: 10.1136/bmj.l5515).

For these patients, “We suggest screening with one of the four screening options: fecal immunochemical test (FIT) every year, FIT every 2 years, a single sigmoidoscopy, or a single colonoscopy,” wrote Dr. Helsingen of the University of Oslo, and her team.

She chaired a 22-member international panel that developed a collaborative effort from the MAGIC research and innovation program as a part of the BMJ Rapid Recommendations project. The team reviewed 12 research papers comprising almost 1.4 million patients from Denmark, Italy, the Netherlands, Norway, Poland, Spain, Sweden, the United Kingdom, and the United States. Follow-up ranged from 0 to 19.5 years for colorectal cancer incidence and up to 30 years for mortality.

Because of the dearth of relevant data in some studies, however, the projected outcomes had to be simulated, with benefits and harms calculations based on 100% screening adherence. However, the team noted, it’s impossible to achieve complete adherence. Most studies of colorectal screening don’t exceed a 50% adherence level.

“All the modeling data are of low certainty. It is a useful indication, but there is a high chance that new evidence will show a smaller or larger benefit, which in turn may alter these recommendations.”

Compared with no screening, all four screening models reduced the risk of colorectal cancer mortality to a similar level.

• FIT every year, 59%.
• FIT every 2 years, 50%.
• Single sigmoidoscopy, 52%.
• Single colonoscopy, 67%.

Screening had less of an impact on reducing the incidence of colorectal cancer:

• FIT every 2 years, 0.05%.
• FIT every year, 0.15%.
• Single sigmoidoscopy, 27%.
• Single colonoscopy, 34%.

The panel also assessed potential harms. Among almost 1 million patients, the colonoscopy-related mortality rate was 0.03 per 1,000 procedures. The perforation rate was 0.8 per 1,000 colonoscopies after a positive fecal test, and 1.4 per 1,000 screened with sigmoidoscopy. The bleeding rate was 1.9 per 1,000 colonoscopies performed after a positive fecal test, and 3-4 per 1,000 screened with sigmoidoscopy.

Successful implementation of these recommendations hinges on accurate risk assessment, however. The team recommended the QCancer platform as “one of the best performing models for both men and women.”

The calculator includes age, sex, ethnicity, smoking status, alcohol use, family history of gastrointestinal cancer, personal history of other cancers, diabetes, ulcerative colitis, colonic polyps, and body mass index.

“We suggest this model because it is available as an online calculator; includes only risk factors available in routine health care; has been validated in a population separate from the derivation population; has reasonable discriminatory ability; and has a good fit between predicted and observed outcomes. In addition, it is the only online risk calculator we know of that predicts risk over a 15-year time horizon.”

The team stressed that their recommendations can’t be applied to all patients. Because evidence for both screening recommendations was weak – largely because of the dearth of supporting data – patients and physicians should cocreate a personalized screening plan.

“Several factors influence individuals’ decisions whether to be screened, even when they are presented with the same information,” the authors said. These include variation in an individual’s values and preferences, a close balance of benefits versus harms and burdens, and personal preference.

“Some individuals may value a minimally invasive test such as FIT, and the possibility of invasive screening with colonoscopy might put them off screening altogether. Those who most value preventing colorectal cancer or avoiding repeated testing are likely to choose sigmoidoscopy or colonoscopy.”

The authors had no financial conflicts of interest.

SOURCE: BMJ 2019;367:l5515. doi: 10.1136/bmj.l5515.

Elderly patients with colorectal cancer have better perioperative outcomes than do those who undergo open surgery, although in this sample, long-term outcomes are similar, Sicheng Zhou, and colleagues wrote in BMC Surgery.

HRAUN/Getty Images

“Laparoscopic surgery showed better results than the open surgery in short-term outcomes,” said Dr. Zhou of the Chinese Academy of Medical Sciences, and coauthors. “[Carcinoembryonic antigen] level, III/IV stage, and perineural invasion were all reliable predictors of overall survival and disease-free survival for the treatment of laparoscopic surgery and open surgery for elderly Chinese patients over 80 years old with colorectal cancer.”

The study comprised 313 patients aged 80 years or older who underwent surgery for colorectal cancer. The group was equally divided between those who had laparoscopic and open surgery. They were matched 1:1, for a total of 93 pairs included. The patients’ mean age was 82 years. Medical comorbidities were present in about 63%. The tumor was more likely to present in the in the rectum and right colon (about 34% each). The next most common disease site was the sigmoid colon (22%).

Most tumors were stage III (58%), and II (about 30%). About 70% were moderately differentiated and 20% poorly differentiated. Carcinoembryonic antigen was greater than 5 ng/mL in about three-fourths of the group, and higher in the reminder.

Surgery duration was somewhat shorter in the open group but not significantly so. However, intraoperative complications were higher in the open group, including transfusion (22.6% vs. 16% and blood loss 50.9 vs. 108 mL). There was a lower occurrence of postoperative complications (10.8% vs. 26.9%) in the laparoscopic group.

Intraoperative complications occurred in only one patient, who was in the laparoscopic group, but perioperative complications were significantly more common in the open group (17.2% vs. 6.5%). In the open group these included wound infection (9.7%), followed by ileus (5.4%), anastomosis leakage (4.3%), and delayed gastric emptying (4.3%). In the laparoscopic group, the most common morbidities were anastomosis leakage (2.2%), ileus (2.2%) and pneumonia (2.2%).

The number of retrieved lymph nodes was also significantly higher in the laparoscopic group (20 vs. 17).

Yet, despite the short-term perioperative advantages of the laparoscopic approach in elderly patients, the 3- and 5-year overall and disease-free survival rates were not significantly different in these groups. The investigators concluded “it is noteworthy that the 3-year and 5-year [overall survival] rates, and 3- year and 5-year [disease-free survival] rates of patients in the laparoscopic group were generally higher than the open group. The 5-year [disease-free survival] rate in the laparoscopic group was even higher than that in the open group by more than 10%. This difference might be due to the difference in the number of dissected lymph nodes between the open group and the laparoscopic group. Hence, although there was no significant difference in survival outcomes between the two surgical methods, the laparoscopic surgery in elderly patients with colorectal cancer might achieve better survival outcomes than the open surgery.”

The authors declared that they had no competing interests. This work was supported by the Beijing Hope Run Special Fund of Cancer Foundation of China and Capital Health Research and Development.

[email protected]

SOURCE: Zhou S et al. BMC Surg. 2019;19:137. doi: 10.1186/s12893-019-0596-3.

Elderly patients with colorectal cancer have better perioperative outcomes than do those who undergo open surgery, although in this sample, long-term outcomes are similar, Sicheng Zhou, and colleagues wrote in BMC Surgery.

HRAUN/Getty Images

“Laparoscopic surgery showed better results than the open surgery in short-term outcomes,” said Dr. Zhou of the Chinese Academy of Medical Sciences, and coauthors. “[Carcinoembryonic antigen] level, III/IV stage, and perineural invasion were all reliable predictors of overall survival and disease-free survival for the treatment of laparoscopic surgery and open surgery for elderly Chinese patients over 80 years old with colorectal cancer.”

The study comprised 313 patients aged 80 years or older who underwent surgery for colorectal cancer. The group was equally divided between those who had laparoscopic and open surgery. They were matched 1:1, for a total of 93 pairs included. The patients’ mean age was 82 years. Medical comorbidities were present in about 63%. The tumor was more likely to present in the in the rectum and right colon (about 34% each). The next most common disease site was the sigmoid colon (22%).

Most tumors were stage III (58%), and II (about 30%). About 70% were moderately differentiated and 20% poorly differentiated. Carcinoembryonic antigen was greater than 5 ng/mL in about three-fourths of the group, and higher in the reminder.

Surgery duration was somewhat shorter in the open group but not significantly so. However, intraoperative complications were higher in the open group, including transfusion (22.6% vs. 16% and blood loss 50.9 vs. 108 mL). There was a lower occurrence of postoperative complications (10.8% vs. 26.9%) in the laparoscopic group.

Intraoperative complications occurred in only one patient, who was in the laparoscopic group, but perioperative complications were significantly more common in the open group (17.2% vs. 6.5%). In the open group these included wound infection (9.7%), followed by ileus (5.4%), anastomosis leakage (4.3%), and delayed gastric emptying (4.3%). In the laparoscopic group, the most common morbidities were anastomosis leakage (2.2%), ileus (2.2%) and pneumonia (2.2%).

The number of retrieved lymph nodes was also significantly higher in the laparoscopic group (20 vs. 17).

Yet, despite the short-term perioperative advantages of the laparoscopic approach in elderly patients, the 3- and 5-year overall and disease-free survival rates were not significantly different in these groups. The investigators concluded “it is noteworthy that the 3-year and 5-year [overall survival] rates, and 3- year and 5-year [disease-free survival] rates of patients in the laparoscopic group were generally higher than the open group. The 5-year [disease-free survival] rate in the laparoscopic group was even higher than that in the open group by more than 10%. This difference might be due to the difference in the number of dissected lymph nodes between the open group and the laparoscopic group. Hence, although there was no significant difference in survival outcomes between the two surgical methods, the laparoscopic surgery in elderly patients with colorectal cancer might achieve better survival outcomes than the open surgery.”

The authors declared that they had no competing interests. This work was supported by the Beijing Hope Run Special Fund of Cancer Foundation of China and Capital Health Research and Development.

[email protected]

SOURCE: Zhou S et al. BMC Surg. 2019;19:137. doi: 10.1186/s12893-019-0596-3.

Elderly patients with colorectal cancer have better perioperative outcomes than do those who undergo open surgery, although in this sample, long-term outcomes are similar, Sicheng Zhou, and colleagues wrote in BMC Surgery.

HRAUN/Getty Images

“Laparoscopic surgery showed better results than the open surgery in short-term outcomes,” said Dr. Zhou of the Chinese Academy of Medical Sciences, and coauthors. “[Carcinoembryonic antigen] level, III/IV stage, and perineural invasion were all reliable predictors of overall survival and disease-free survival for the treatment of laparoscopic surgery and open surgery for elderly Chinese patients over 80 years old with colorectal cancer.”

The study comprised 313 patients aged 80 years or older who underwent surgery for colorectal cancer. The group was equally divided between those who had laparoscopic and open surgery. They were matched 1:1, for a total of 93 pairs included. The patients’ mean age was 82 years. Medical comorbidities were present in about 63%. The tumor was more likely to present in the in the rectum and right colon (about 34% each). The next most common disease site was the sigmoid colon (22%).

Most tumors were stage III (58%), and II (about 30%). About 70% were moderately differentiated and 20% poorly differentiated. Carcinoembryonic antigen was greater than 5 ng/mL in about three-fourths of the group, and higher in the reminder.

Surgery duration was somewhat shorter in the open group but not significantly so. However, intraoperative complications were higher in the open group, including transfusion (22.6% vs. 16% and blood loss 50.9 vs. 108 mL). There was a lower occurrence of postoperative complications (10.8% vs. 26.9%) in the laparoscopic group.

Intraoperative complications occurred in only one patient, who was in the laparoscopic group, but perioperative complications were significantly more common in the open group (17.2% vs. 6.5%). In the open group these included wound infection (9.7%), followed by ileus (5.4%), anastomosis leakage (4.3%), and delayed gastric emptying (4.3%). In the laparoscopic group, the most common morbidities were anastomosis leakage (2.2%), ileus (2.2%) and pneumonia (2.2%).

The number of retrieved lymph nodes was also significantly higher in the laparoscopic group (20 vs. 17).

Yet, despite the short-term perioperative advantages of the laparoscopic approach in elderly patients, the 3- and 5-year overall and disease-free survival rates were not significantly different in these groups. The investigators concluded “it is noteworthy that the 3-year and 5-year [overall survival] rates, and 3- year and 5-year [disease-free survival] rates of patients in the laparoscopic group were generally higher than the open group. The 5-year [disease-free survival] rate in the laparoscopic group was even higher than that in the open group by more than 10%. This difference might be due to the difference in the number of dissected lymph nodes between the open group and the laparoscopic group. Hence, although there was no significant difference in survival outcomes between the two surgical methods, the laparoscopic surgery in elderly patients with colorectal cancer might achieve better survival outcomes than the open surgery.”

The authors declared that they had no competing interests. This work was supported by the Beijing Hope Run Special Fund of Cancer Foundation of China and Capital Health Research and Development.

[email protected]

SOURCE: Zhou S et al. BMC Surg. 2019;19:137. doi: 10.1186/s12893-019-0596-3.

Artificial intelligence (AI), first described in 1956, encompasses the field of computer science in which machines are trained to learn from experience. The term was popularized by the 1956 Dartmouth College Summer Research Project on Artificial Intelligence.¹ The field of AI is rapidly growing and has the potential to affect many aspects of our lives. The emerging importance of AI is demonstrated by a February 2019 executive order that launched the American AI Initiative, allocating resources and funding for AI development.² The executive order stresses the potential impact of AI in the health care field, including its potential utility to diagnose disease. Federal agencies were directed to invest in AI research and development to promote rapid breakthroughs in AI technology that may impact multiple areas of society.

Machine learning (ML), a subset of AI, was defined in 1959 by Arthur Samuel and is achieved by employing mathematic models to compute sample data sets.³ Originating from statistical linear models, neural networks were conceived to accomplish these tasks.⁴ These pioneering scientific achievements led to recent developments of deep neural networks. These models are developed to recognize patterns and achieve complex computational tasks within a matter of minutes, often far exceeding human ability.⁵ ML can increase efficiency with decreased computation time, high precision, and recall when compared with that of human decision making.⁶

ML has the potential for numerous applications in the health care field.^7-9 One promising application is in the field of anatomic pathology. ML allows representative images to be used to train a computer to recognize patterns from labeled photographs. Based on a set of images selected to represent a specific tissue or disease process, the computer can be trained to evaluate and recognize new and unique images from patients and render a diagnosis.¹⁰ Prior to modern ML models, users would have to import many thousands of training images to produce algorithms that could recognize patterns with high accuracy. Modern ML algorithms allow for a model known as transfer learning, such that far fewer images are required for training.^11-13

Two novel ML platforms available for public use are offered through Google (Mountain View, CA) and Apple (Cupertino, CA).^14,15 They each offer a user-friendly interface with minimal experience required in computer science. Google AutoML uses ML via cloud services to store and retrieve data with ease. No coding knowledge is required. The Apple Create ML Module provides computer-based ML, requiring only a few lines of code.

The Veterans Health Administration (VHA) is the largest single health care system in the US, and nearly 50 000 cancer cases are diagnosed at the VHA annually.¹⁶ Cancers of the lung and colon are among the most common sources of invasive cancer and are the 2 most common causes of cancer deaths in America.¹⁶ We have previously reported using Apple ML in detecting non-small cell lung cancers (NSCLCs), including adenocarcinomas and squamous cell carcinomas (SCCs); and colon cancers with accuracy.^17,18 In the present study, we expand on these findings by comparing Apple and Google ML platforms in a variety of common pathologic scenarios in veteran patients. Using limited training data, both programs are compared for precision and recall in differentiating conditions involving lung and colon pathology.

In the first 4 experiments, we evaluated the ability of the platforms to differentiate normal lung tissue from cancerous lung tissue, to distinguish lung adenocarcinoma from SCC, and to differentiate colon adenocarcinoma from normal colon tissue. Next, cases of colon adenocarcinoma were assessed to determine whether the presence or absence of the KRAS proto-oncogene could be determined histologically using the AI platforms. KRAS is found in a variety of cancers, including about 40% of colon adenocarcinomas.¹⁹ For colon cancers, the presence or absence of the mutation in KRAS has important implications for patients as it determines whether the tumor will respond to specific chemotherapy agents.²⁰ The presence of the KRAS gene is currently determined by complex molecular testing of tumor tissue.²¹ However, we assessed the potential of ML to determine whether the mutation is present by computerized morphologic analysis alone. Our last experiment examined the ability of the Apple and Google platforms to differentiate between adenocarcinomas of lung origin vs colon origin. This has potential utility in determining the site of origin of metastatic carcinoma.²²

Methods

Fifty cases of lung SCC, 50 cases of lung adenocarcinoma, and 50 cases of colon adenocarcinoma were randomly retrieved from our molecular database. Twenty-five colon adenocarcinoma cases were positive for mutation in KRAS, while 25 cases were negative for mutation in KRAS. Seven hundred fifty total images of lung tissue (250 benign lung tissue, 250 lung adenocarcinomas, and 250 lung SCCs) and 500 total images of colon tissue (250 benign colon tissue and 250 colon adenocarcinoma) were obtained using a Leica Microscope MC190 HD Camera (Wetzlar, Germany) connected to an Olympus BX41 microscope (Center Valley, PA) and the Leica Acquire 9072 software for Apple computers. All the images were captured at a resolution of 1024 x 768 pixels using a 60x dry objective. Lung tissue images were captured and saved on a 2012 Apple MacBook Pro computer, and colon images were captured and saved on a 2011 Apple iMac computer. Both computers were running macOS v10.13.

Creating Image Classifier Models Using Apple Create ML

Apple Create ML is a suite of products that use various tools to create and train custom ML models on Apple computers.¹⁵ The suite contains many features, including image classification to train a ML model to classify images, natural language processing to classify natural language text, and tabular data to train models that deal with labeling information or estimating new quantities. We used Create ML Image Classification to create image classifier models for our project (Appendix A).

Creating ML Modules Using Google Cloud AutoML Vision Beta

Google Cloud AutoML is a suite of machine learning products, including AutoML Vision, AutoML Natural Language and AutoML Translation.¹⁴ All Cloud AutoML machine learning products were in beta version at the time of experimentation. We used Cloud AutoML Vision beta to create ML modules for our project. Unlike Apple Create ML, which is run on a local Apple computer, the Google Cloud AutoML is run online using a Google Cloud account. There are no minimum specifications requirements for the local computer since it is using the cloud-based architecture (Appendix B).

Experiment 1

We compared Apple Create ML Image Classifier and Google AutoML Vision in their ability to detect and subclassify NSCLC based on the histopathologic images. We created 3 classes of images (250 images each): benign lung tissue, lung adenocarcinoma, and lung SCC.

Experiment 2

We compared Apple Create ML Image Classifier and Google AutoML Vision in their ability to differentiate between normal lung tissue and NSCLC histopathologic images with 50/50 mixture of lung adenocarcinoma and lung SCC. We created 2 classes of images (250 images each): benign lung tissue and lung NSCLC.

Experiment 3

We compared Apple Create ML Image Classifier and Google AutoML Vision in their ability to differentiate between lung adenocarcinoma and lung SCC histopathologic images. We created 2 classes of images (250 images each): adenocarcinoma and SCC.

Experiment 4

We compared Apple Create ML Image Classifier and Google AutoML Vision in their ability to detect colon cancer histopathologic images regardless of mutation in KRAS status. We created 2 classes of images (250 images each): benign colon tissue and colon adenocarcinoma.

Experiment 5

We compared Apple Create ML Image Classifier and Google AutoML Vision in their ability to differentiate between colon adenocarcinoma with mutations in KRAS and colon adenocarcinoma without the mutation in KRAS histopathologic images. We created 2 classes of images (125 images each): colon adenocarcinoma cases with mutation in KRAS and colon adenocarcinoma cases without the mutation in KRAS.

Experiment 6

We compared Apple Create ML Image Classifier and Google AutoML Vision in their ability to differentiate between lung adenocarcinoma and colon adenocarcinoma histopathologic images. We created 2 classes of images (250 images each): colon adenocarcinoma lung adenocarcinoma.

Results

Twelve machine learning models were created in 6 experiments using the Apple Create ML and the Google AutoML (Table). To investigate recall and precision differences between the Apple and the Google ML algorithms, we performed 2-tailed distribution, paired t tests. No statistically significant differences were found (P = .52 for recall and .60 for precision).

Overall, each model performed well in distinguishing between normal and neoplastic tissue for both lung and colon cancers. In subclassifying NSCLC into adenocarcinoma and SCC, the models were shown to have high levels of precision and recall. The models also were successful in distinguishing between lung and colonic origin of adenocarcinoma (Figures 1-4). However, both systems had trouble discerning colon adenocarcinoma with mutations in KRAS from adenocarcinoma without mutations in KRAS.

Discussion

Image classifier models using ML algorithms hold a promising future to revolutionize the health care field. ML products, such as those modules offered by Apple and Google, are easy to use and have a simple graphic user interface to allow individuals to train models to perform humanlike tasks in real time. In our experiments, we compared multiple algorithms to determine their ability to differentiate and subclassify histopathologic images with high precision and recall using common scenarios in treating veteran patients.

Analysis of the results revealed high precision and recall values illustrating the models’ ability to differentiate and detect benign lung tissue from lung SCC and lung adenocarcinoma in ML model 1, benign lung from NSCLC carcinoma in ML model 2, and benign colon from colonic adenocarcinoma in ML model 4. In ML model 3 and 6, both ML algorithms performed at a high level to differentiate lung SCC from lung adenocarcinoma and lung adenocarcinoma from colonic adenocarcinoma, respectively. Of note, ML model 5 had the lowest precision and recall values across both algorithms demonstrating the models’ limited utility in predicting molecular profiles, such as mutations in KRAS as tested here. This is not surprising as pathologists currently require complex molecular tests to detect mutations in KRAS reliably in colon cancer.

Both modules require minimal programming experience and are easy to use. In our comparison, we demonstrated critical distinguishing characteristics that differentiate the 2 products.

Apple Create ML image classifier is available for use on local Mac computers that use Xcode version 10 and macOS 10.14 or later, with just 3 lines of code required to perform computations. Although this product is limited to Apple computers, it is free to use, and images are stored on the computer hard drive. Of unique significance on the Apple system platform, images can be augmented to alter their appearance to enhance model training. For example, imported images can be cropped, rotated, blurred, and flipped, in order to optimize the model’s training abilities to recognize test images and perform pattern recognition. This feature is not as readily available on the Google platform. Apple Create ML Image classifier’s default training set consists of 75% of total imported images with 5% of the total images being randomly used as a validation set. The remaining 20% of images comprise the testing set. The module’s computational analysis to train the model is achieved in about 2 minutes on average. The score threshold is set at 50% and cannot be manipulated for each image class as in Google AutoML Vision.

Google AutoML Vision is open and can be accessed from many devices. It stores images on remote Google servers but requires computing fees after a $300 credit for 12 months. On AutoML Vision, random 80% of the total images are used in the training set, 10% are used in the validation set, and 10% are used in the testing set. It is important to highlight the different percentages used in the default settings on the respective modules. The time to train the Google AutoML Vision with default computational power is longer on average than Apple Create ML, with about 8 minutes required to train the machine learning module. However, it is possible to choose more computational power for an additional fee and decrease module training time. The user will receive e-mail alerts when the computer time begins and is completed. The computation time is calculated by subtracting the time of the initial e-mail from the final e-mail.

Based on our calculations, we determined there was no significant difference between the 2 machine learning algorithms tested at the default settings with recall and precision values obtained. These findings demonstrate the promise of using a ML algorithm to assist in the performance of human tasks and behaviors, specifically the diagnosis of histopathologic images. These results have numerous potential uses in clinical medicine. ML algorithms have been successfully applied to diagnostic and prognostic endeavors in pathology,^23-28 dermatology,^29-31 ophthalmology,³² cardiology,³³ and radiology.^34-36

Pathologists often use additional tests, such as special staining of tissues or molecular tests, to assist with accurate classification of tumors. ML platforms offer the potential of an additional tool for pathologists to use along with human microscopic interpretation.^37,38 In addition, the number of pathologists in the US is dramatically decreasing, and many other countries have marked physician shortages, especially in fields of specialized training such as pathology.^39-42 These models could readily assist physicians in underserved countries and impact shortages of pathologists elsewhere by providing more specific diagnoses in an expedited manner.⁴³

Finally, although we have explored the application of these platforms in common cancer scenarios, great potential exists to use similar techniques in the detection of other conditions. These include the potential for classification and risk assessment of precancerous lesions, infectious processes in tissue (eg, detection of tuberculosis or malaria),^24,44 inflammatory conditions (eg, arthritis subtypes, gout),⁴⁵ blood disorders (eg, abnormal blood cell morphology),⁴⁶ and many others. The potential of these technologies to improve health care delivery to veteran patients seems to be limited only by the imagination of the user.⁴⁷

Regarding the limited effectiveness in determining the presence or absence of mutations in KRAS for colon adenocarcinoma, it is mentioned that currently pathologists rely on complex molecular tests to detect the mutations at the DNA level.²¹ It is possible that the use of more extensive training data sets may improve recall and precision in cases such as these and warrants further study. Our experiments were limited to the stipulations placed by the free trial software agreements; no costs were expended to use the algorithms, though an Apple computer was required.

Conclusion

We have demonstrated the successful application of 2 readily available ML platforms in providing diagnostic guidance in differentiation between common cancer conditions in veteran patient populations. Although both platforms performed very well with no statistically significant differences in results, some distinctions are worth noting. Apple Create ML can be used on local computers but is limited to an Apple operating system. Google AutoML is not platform-specific but runs only via Google Cloud with associated computational fees. Using these readily available models, we demonstrated the vast potential of AI in diagnostic pathology. The application of AI to clinical medicine remains in the very early stages. The VA is uniquely poised to provide leadership as AI technologies will continue to dramatically change the future of health care, both in veteran and nonveteran patients nationwide.

Acknowledgments

The authors thank Paul Borkowski for his constructive criticism and proofreading of this manuscript. This material is the result of work supported with resources and the use of facilities at the James A. Haley Veterans’ Hospital.

Artificial intelligence (AI), first described in 1956, encompasses the field of computer science in which machines are trained to learn from experience. The term was popularized by the 1956 Dartmouth College Summer Research Project on Artificial Intelligence.¹ The field of AI is rapidly growing and has the potential to affect many aspects of our lives. The emerging importance of AI is demonstrated by a February 2019 executive order that launched the American AI Initiative, allocating resources and funding for AI development.² The executive order stresses the potential impact of AI in the health care field, including its potential utility to diagnose disease. Federal agencies were directed to invest in AI research and development to promote rapid breakthroughs in AI technology that may impact multiple areas of society.

Machine learning (ML), a subset of AI, was defined in 1959 by Arthur Samuel and is achieved by employing mathematic models to compute sample data sets.³ Originating from statistical linear models, neural networks were conceived to accomplish these tasks.⁴ These pioneering scientific achievements led to recent developments of deep neural networks. These models are developed to recognize patterns and achieve complex computational tasks within a matter of minutes, often far exceeding human ability.⁵ ML can increase efficiency with decreased computation time, high precision, and recall when compared with that of human decision making.⁶

ML has the potential for numerous applications in the health care field.^7-9 One promising application is in the field of anatomic pathology. ML allows representative images to be used to train a computer to recognize patterns from labeled photographs. Based on a set of images selected to represent a specific tissue or disease process, the computer can be trained to evaluate and recognize new and unique images from patients and render a diagnosis.¹⁰ Prior to modern ML models, users would have to import many thousands of training images to produce algorithms that could recognize patterns with high accuracy. Modern ML algorithms allow for a model known as transfer learning, such that far fewer images are required for training.^11-13

Two novel ML platforms available for public use are offered through Google (Mountain View, CA) and Apple (Cupertino, CA).^14,15 They each offer a user-friendly interface with minimal experience required in computer science. Google AutoML uses ML via cloud services to store and retrieve data with ease. No coding knowledge is required. The Apple Create ML Module provides computer-based ML, requiring only a few lines of code.

The Veterans Health Administration (VHA) is the largest single health care system in the US, and nearly 50 000 cancer cases are diagnosed at the VHA annually.¹⁶ Cancers of the lung and colon are among the most common sources of invasive cancer and are the 2 most common causes of cancer deaths in America.¹⁶ We have previously reported using Apple ML in detecting non-small cell lung cancers (NSCLCs), including adenocarcinomas and squamous cell carcinomas (SCCs); and colon cancers with accuracy.^17,18 In the present study, we expand on these findings by comparing Apple and Google ML platforms in a variety of common pathologic scenarios in veteran patients. Using limited training data, both programs are compared for precision and recall in differentiating conditions involving lung and colon pathology.

In the first 4 experiments, we evaluated the ability of the platforms to differentiate normal lung tissue from cancerous lung tissue, to distinguish lung adenocarcinoma from SCC, and to differentiate colon adenocarcinoma from normal colon tissue. Next, cases of colon adenocarcinoma were assessed to determine whether the presence or absence of the KRAS proto-oncogene could be determined histologically using the AI platforms. KRAS is found in a variety of cancers, including about 40% of colon adenocarcinomas.¹⁹ For colon cancers, the presence or absence of the mutation in KRAS has important implications for patients as it determines whether the tumor will respond to specific chemotherapy agents.²⁰ The presence of the KRAS gene is currently determined by complex molecular testing of tumor tissue.²¹ However, we assessed the potential of ML to determine whether the mutation is present by computerized morphologic analysis alone. Our last experiment examined the ability of the Apple and Google platforms to differentiate between adenocarcinomas of lung origin vs colon origin. This has potential utility in determining the site of origin of metastatic carcinoma.²²

Methods

Fifty cases of lung SCC, 50 cases of lung adenocarcinoma, and 50 cases of colon adenocarcinoma were randomly retrieved from our molecular database. Twenty-five colon adenocarcinoma cases were positive for mutation in KRAS, while 25 cases were negative for mutation in KRAS. Seven hundred fifty total images of lung tissue (250 benign lung tissue, 250 lung adenocarcinomas, and 250 lung SCCs) and 500 total images of colon tissue (250 benign colon tissue and 250 colon adenocarcinoma) were obtained using a Leica Microscope MC190 HD Camera (Wetzlar, Germany) connected to an Olympus BX41 microscope (Center Valley, PA) and the Leica Acquire 9072 software for Apple computers. All the images were captured at a resolution of 1024 x 768 pixels using a 60x dry objective. Lung tissue images were captured and saved on a 2012 Apple MacBook Pro computer, and colon images were captured and saved on a 2011 Apple iMac computer. Both computers were running macOS v10.13.

Creating Image Classifier Models Using Apple Create ML

Apple Create ML is a suite of products that use various tools to create and train custom ML models on Apple computers.¹⁵ The suite contains many features, including image classification to train a ML model to classify images, natural language processing to classify natural language text, and tabular data to train models that deal with labeling information or estimating new quantities. We used Create ML Image Classification to create image classifier models for our project (Appendix A).

Creating ML Modules Using Google Cloud AutoML Vision Beta

Google Cloud AutoML is a suite of machine learning products, including AutoML Vision, AutoML Natural Language and AutoML Translation.¹⁴ All Cloud AutoML machine learning products were in beta version at the time of experimentation. We used Cloud AutoML Vision beta to create ML modules for our project. Unlike Apple Create ML, which is run on a local Apple computer, the Google Cloud AutoML is run online using a Google Cloud account. There are no minimum specifications requirements for the local computer since it is using the cloud-based architecture (Appendix B).

Experiment 1

We compared Apple Create ML Image Classifier and Google AutoML Vision in their ability to detect and subclassify NSCLC based on the histopathologic images. We created 3 classes of images (250 images each): benign lung tissue, lung adenocarcinoma, and lung SCC.

Experiment 2

We compared Apple Create ML Image Classifier and Google AutoML Vision in their ability to differentiate between normal lung tissue and NSCLC histopathologic images with 50/50 mixture of lung adenocarcinoma and lung SCC. We created 2 classes of images (250 images each): benign lung tissue and lung NSCLC.

Experiment 3

We compared Apple Create ML Image Classifier and Google AutoML Vision in their ability to differentiate between lung adenocarcinoma and lung SCC histopathologic images. We created 2 classes of images (250 images each): adenocarcinoma and SCC.

Experiment 4

We compared Apple Create ML Image Classifier and Google AutoML Vision in their ability to detect colon cancer histopathologic images regardless of mutation in KRAS status. We created 2 classes of images (250 images each): benign colon tissue and colon adenocarcinoma.

Experiment 5

We compared Apple Create ML Image Classifier and Google AutoML Vision in their ability to differentiate between colon adenocarcinoma with mutations in KRAS and colon adenocarcinoma without the mutation in KRAS histopathologic images. We created 2 classes of images (125 images each): colon adenocarcinoma cases with mutation in KRAS and colon adenocarcinoma cases without the mutation in KRAS.

Experiment 6

We compared Apple Create ML Image Classifier and Google AutoML Vision in their ability to differentiate between lung adenocarcinoma and colon adenocarcinoma histopathologic images. We created 2 classes of images (250 images each): colon adenocarcinoma lung adenocarcinoma.

Results

Twelve machine learning models were created in 6 experiments using the Apple Create ML and the Google AutoML (Table). To investigate recall and precision differences between the Apple and the Google ML algorithms, we performed 2-tailed distribution, paired t tests. No statistically significant differences were found (P = .52 for recall and .60 for precision).

Overall, each model performed well in distinguishing between normal and neoplastic tissue for both lung and colon cancers. In subclassifying NSCLC into adenocarcinoma and SCC, the models were shown to have high levels of precision and recall. The models also were successful in distinguishing between lung and colonic origin of adenocarcinoma (Figures 1-4). However, both systems had trouble discerning colon adenocarcinoma with mutations in KRAS from adenocarcinoma without mutations in KRAS.

Discussion

Image classifier models using ML algorithms hold a promising future to revolutionize the health care field. ML products, such as those modules offered by Apple and Google, are easy to use and have a simple graphic user interface to allow individuals to train models to perform humanlike tasks in real time. In our experiments, we compared multiple algorithms to determine their ability to differentiate and subclassify histopathologic images with high precision and recall using common scenarios in treating veteran patients.

Analysis of the results revealed high precision and recall values illustrating the models’ ability to differentiate and detect benign lung tissue from lung SCC and lung adenocarcinoma in ML model 1, benign lung from NSCLC carcinoma in ML model 2, and benign colon from colonic adenocarcinoma in ML model 4. In ML model 3 and 6, both ML algorithms performed at a high level to differentiate lung SCC from lung adenocarcinoma and lung adenocarcinoma from colonic adenocarcinoma, respectively. Of note, ML model 5 had the lowest precision and recall values across both algorithms demonstrating the models’ limited utility in predicting molecular profiles, such as mutations in KRAS as tested here. This is not surprising as pathologists currently require complex molecular tests to detect mutations in KRAS reliably in colon cancer.

Both modules require minimal programming experience and are easy to use. In our comparison, we demonstrated critical distinguishing characteristics that differentiate the 2 products.

Apple Create ML image classifier is available for use on local Mac computers that use Xcode version 10 and macOS 10.14 or later, with just 3 lines of code required to perform computations. Although this product is limited to Apple computers, it is free to use, and images are stored on the computer hard drive. Of unique significance on the Apple system platform, images can be augmented to alter their appearance to enhance model training. For example, imported images can be cropped, rotated, blurred, and flipped, in order to optimize the model’s training abilities to recognize test images and perform pattern recognition. This feature is not as readily available on the Google platform. Apple Create ML Image classifier’s default training set consists of 75% of total imported images with 5% of the total images being randomly used as a validation set. The remaining 20% of images comprise the testing set. The module’s computational analysis to train the model is achieved in about 2 minutes on average. The score threshold is set at 50% and cannot be manipulated for each image class as in Google AutoML Vision.

Google AutoML Vision is open and can be accessed from many devices. It stores images on remote Google servers but requires computing fees after a $300 credit for 12 months. On AutoML Vision, random 80% of the total images are used in the training set, 10% are used in the validation set, and 10% are used in the testing set. It is important to highlight the different percentages used in the default settings on the respective modules. The time to train the Google AutoML Vision with default computational power is longer on average than Apple Create ML, with about 8 minutes required to train the machine learning module. However, it is possible to choose more computational power for an additional fee and decrease module training time. The user will receive e-mail alerts when the computer time begins and is completed. The computation time is calculated by subtracting the time of the initial e-mail from the final e-mail.

Based on our calculations, we determined there was no significant difference between the 2 machine learning algorithms tested at the default settings with recall and precision values obtained. These findings demonstrate the promise of using a ML algorithm to assist in the performance of human tasks and behaviors, specifically the diagnosis of histopathologic images. These results have numerous potential uses in clinical medicine. ML algorithms have been successfully applied to diagnostic and prognostic endeavors in pathology,^23-28 dermatology,^29-31 ophthalmology,³² cardiology,³³ and radiology.^34-36

Pathologists often use additional tests, such as special staining of tissues or molecular tests, to assist with accurate classification of tumors. ML platforms offer the potential of an additional tool for pathologists to use along with human microscopic interpretation.^37,38 In addition, the number of pathologists in the US is dramatically decreasing, and many other countries have marked physician shortages, especially in fields of specialized training such as pathology.^39-42 These models could readily assist physicians in underserved countries and impact shortages of pathologists elsewhere by providing more specific diagnoses in an expedited manner.⁴³

Finally, although we have explored the application of these platforms in common cancer scenarios, great potential exists to use similar techniques in the detection of other conditions. These include the potential for classification and risk assessment of precancerous lesions, infectious processes in tissue (eg, detection of tuberculosis or malaria),^24,44 inflammatory conditions (eg, arthritis subtypes, gout),⁴⁵ blood disorders (eg, abnormal blood cell morphology),⁴⁶ and many others. The potential of these technologies to improve health care delivery to veteran patients seems to be limited only by the imagination of the user.⁴⁷

Regarding the limited effectiveness in determining the presence or absence of mutations in KRAS for colon adenocarcinoma, it is mentioned that currently pathologists rely on complex molecular tests to detect the mutations at the DNA level.²¹ It is possible that the use of more extensive training data sets may improve recall and precision in cases such as these and warrants further study. Our experiments were limited to the stipulations placed by the free trial software agreements; no costs were expended to use the algorithms, though an Apple computer was required.

Conclusion

We have demonstrated the successful application of 2 readily available ML platforms in providing diagnostic guidance in differentiation between common cancer conditions in veteran patient populations. Although both platforms performed very well with no statistically significant differences in results, some distinctions are worth noting. Apple Create ML can be used on local computers but is limited to an Apple operating system. Google AutoML is not platform-specific but runs only via Google Cloud with associated computational fees. Using these readily available models, we demonstrated the vast potential of AI in diagnostic pathology. The application of AI to clinical medicine remains in the very early stages. The VA is uniquely poised to provide leadership as AI technologies will continue to dramatically change the future of health care, both in veteran and nonveteran patients nationwide.

Acknowledgments

The authors thank Paul Borkowski for his constructive criticism and proofreading of this manuscript. This material is the result of work supported with resources and the use of facilities at the James A. Haley Veterans’ Hospital.

Artificial intelligence (AI), first described in 1956, encompasses the field of computer science in which machines are trained to learn from experience. The term was popularized by the 1956 Dartmouth College Summer Research Project on Artificial Intelligence.¹ The field of AI is rapidly growing and has the potential to affect many aspects of our lives. The emerging importance of AI is demonstrated by a February 2019 executive order that launched the American AI Initiative, allocating resources and funding for AI development.² The executive order stresses the potential impact of AI in the health care field, including its potential utility to diagnose disease. Federal agencies were directed to invest in AI research and development to promote rapid breakthroughs in AI technology that may impact multiple areas of society.

Machine learning (ML), a subset of AI, was defined in 1959 by Arthur Samuel and is achieved by employing mathematic models to compute sample data sets.³ Originating from statistical linear models, neural networks were conceived to accomplish these tasks.⁴ These pioneering scientific achievements led to recent developments of deep neural networks. These models are developed to recognize patterns and achieve complex computational tasks within a matter of minutes, often far exceeding human ability.⁵ ML can increase efficiency with decreased computation time, high precision, and recall when compared with that of human decision making.⁶

ML has the potential for numerous applications in the health care field.^7-9 One promising application is in the field of anatomic pathology. ML allows representative images to be used to train a computer to recognize patterns from labeled photographs. Based on a set of images selected to represent a specific tissue or disease process, the computer can be trained to evaluate and recognize new and unique images from patients and render a diagnosis.¹⁰ Prior to modern ML models, users would have to import many thousands of training images to produce algorithms that could recognize patterns with high accuracy. Modern ML algorithms allow for a model known as transfer learning, such that far fewer images are required for training.^11-13

Two novel ML platforms available for public use are offered through Google (Mountain View, CA) and Apple (Cupertino, CA).^14,15 They each offer a user-friendly interface with minimal experience required in computer science. Google AutoML uses ML via cloud services to store and retrieve data with ease. No coding knowledge is required. The Apple Create ML Module provides computer-based ML, requiring only a few lines of code.

The Veterans Health Administration (VHA) is the largest single health care system in the US, and nearly 50 000 cancer cases are diagnosed at the VHA annually.¹⁶ Cancers of the lung and colon are among the most common sources of invasive cancer and are the 2 most common causes of cancer deaths in America.¹⁶ We have previously reported using Apple ML in detecting non-small cell lung cancers (NSCLCs), including adenocarcinomas and squamous cell carcinomas (SCCs); and colon cancers with accuracy.^17,18 In the present study, we expand on these findings by comparing Apple and Google ML platforms in a variety of common pathologic scenarios in veteran patients. Using limited training data, both programs are compared for precision and recall in differentiating conditions involving lung and colon pathology.

In the first 4 experiments, we evaluated the ability of the platforms to differentiate normal lung tissue from cancerous lung tissue, to distinguish lung adenocarcinoma from SCC, and to differentiate colon adenocarcinoma from normal colon tissue. Next, cases of colon adenocarcinoma were assessed to determine whether the presence or absence of the KRAS proto-oncogene could be determined histologically using the AI platforms. KRAS is found in a variety of cancers, including about 40% of colon adenocarcinomas.¹⁹ For colon cancers, the presence or absence of the mutation in KRAS has important implications for patients as it determines whether the tumor will respond to specific chemotherapy agents.²⁰ The presence of the KRAS gene is currently determined by complex molecular testing of tumor tissue.²¹ However, we assessed the potential of ML to determine whether the mutation is present by computerized morphologic analysis alone. Our last experiment examined the ability of the Apple and Google platforms to differentiate between adenocarcinomas of lung origin vs colon origin. This has potential utility in determining the site of origin of metastatic carcinoma.²²

Methods

Fifty cases of lung SCC, 50 cases of lung adenocarcinoma, and 50 cases of colon adenocarcinoma were randomly retrieved from our molecular database. Twenty-five colon adenocarcinoma cases were positive for mutation in KRAS, while 25 cases were negative for mutation in KRAS. Seven hundred fifty total images of lung tissue (250 benign lung tissue, 250 lung adenocarcinomas, and 250 lung SCCs) and 500 total images of colon tissue (250 benign colon tissue and 250 colon adenocarcinoma) were obtained using a Leica Microscope MC190 HD Camera (Wetzlar, Germany) connected to an Olympus BX41 microscope (Center Valley, PA) and the Leica Acquire 9072 software for Apple computers. All the images were captured at a resolution of 1024 x 768 pixels using a 60x dry objective. Lung tissue images were captured and saved on a 2012 Apple MacBook Pro computer, and colon images were captured and saved on a 2011 Apple iMac computer. Both computers were running macOS v10.13.

Creating Image Classifier Models Using Apple Create ML

Apple Create ML is a suite of products that use various tools to create and train custom ML models on Apple computers.¹⁵ The suite contains many features, including image classification to train a ML model to classify images, natural language processing to classify natural language text, and tabular data to train models that deal with labeling information or estimating new quantities. We used Create ML Image Classification to create image classifier models for our project (Appendix A).

Creating ML Modules Using Google Cloud AutoML Vision Beta

Google Cloud AutoML is a suite of machine learning products, including AutoML Vision, AutoML Natural Language and AutoML Translation.¹⁴ All Cloud AutoML machine learning products were in beta version at the time of experimentation. We used Cloud AutoML Vision beta to create ML modules for our project. Unlike Apple Create ML, which is run on a local Apple computer, the Google Cloud AutoML is run online using a Google Cloud account. There are no minimum specifications requirements for the local computer since it is using the cloud-based architecture (Appendix B).

Experiment 1

We compared Apple Create ML Image Classifier and Google AutoML Vision in their ability to detect and subclassify NSCLC based on the histopathologic images. We created 3 classes of images (250 images each): benign lung tissue, lung adenocarcinoma, and lung SCC.

Experiment 2

We compared Apple Create ML Image Classifier and Google AutoML Vision in their ability to differentiate between normal lung tissue and NSCLC histopathologic images with 50/50 mixture of lung adenocarcinoma and lung SCC. We created 2 classes of images (250 images each): benign lung tissue and lung NSCLC.

Experiment 3

We compared Apple Create ML Image Classifier and Google AutoML Vision in their ability to differentiate between lung adenocarcinoma and lung SCC histopathologic images. We created 2 classes of images (250 images each): adenocarcinoma and SCC.

Experiment 4

We compared Apple Create ML Image Classifier and Google AutoML Vision in their ability to detect colon cancer histopathologic images regardless of mutation in KRAS status. We created 2 classes of images (250 images each): benign colon tissue and colon adenocarcinoma.

Experiment 5

We compared Apple Create ML Image Classifier and Google AutoML Vision in their ability to differentiate between colon adenocarcinoma with mutations in KRAS and colon adenocarcinoma without the mutation in KRAS histopathologic images. We created 2 classes of images (125 images each): colon adenocarcinoma cases with mutation in KRAS and colon adenocarcinoma cases without the mutation in KRAS.

Experiment 6

We compared Apple Create ML Image Classifier and Google AutoML Vision in their ability to differentiate between lung adenocarcinoma and colon adenocarcinoma histopathologic images. We created 2 classes of images (250 images each): colon adenocarcinoma lung adenocarcinoma.

Results

Twelve machine learning models were created in 6 experiments using the Apple Create ML and the Google AutoML (Table). To investigate recall and precision differences between the Apple and the Google ML algorithms, we performed 2-tailed distribution, paired t tests. No statistically significant differences were found (P = .52 for recall and .60 for precision).

Overall, each model performed well in distinguishing between normal and neoplastic tissue for both lung and colon cancers. In subclassifying NSCLC into adenocarcinoma and SCC, the models were shown to have high levels of precision and recall. The models also were successful in distinguishing between lung and colonic origin of adenocarcinoma (Figures 1-4). However, both systems had trouble discerning colon adenocarcinoma with mutations in KRAS from adenocarcinoma without mutations in KRAS.

Discussion

Image classifier models using ML algorithms hold a promising future to revolutionize the health care field. ML products, such as those modules offered by Apple and Google, are easy to use and have a simple graphic user interface to allow individuals to train models to perform humanlike tasks in real time. In our experiments, we compared multiple algorithms to determine their ability to differentiate and subclassify histopathologic images with high precision and recall using common scenarios in treating veteran patients.

Analysis of the results revealed high precision and recall values illustrating the models’ ability to differentiate and detect benign lung tissue from lung SCC and lung adenocarcinoma in ML model 1, benign lung from NSCLC carcinoma in ML model 2, and benign colon from colonic adenocarcinoma in ML model 4. In ML model 3 and 6, both ML algorithms performed at a high level to differentiate lung SCC from lung adenocarcinoma and lung adenocarcinoma from colonic adenocarcinoma, respectively. Of note, ML model 5 had the lowest precision and recall values across both algorithms demonstrating the models’ limited utility in predicting molecular profiles, such as mutations in KRAS as tested here. This is not surprising as pathologists currently require complex molecular tests to detect mutations in KRAS reliably in colon cancer.

Both modules require minimal programming experience and are easy to use. In our comparison, we demonstrated critical distinguishing characteristics that differentiate the 2 products.

Apple Create ML image classifier is available for use on local Mac computers that use Xcode version 10 and macOS 10.14 or later, with just 3 lines of code required to perform computations. Although this product is limited to Apple computers, it is free to use, and images are stored on the computer hard drive. Of unique significance on the Apple system platform, images can be augmented to alter their appearance to enhance model training. For example, imported images can be cropped, rotated, blurred, and flipped, in order to optimize the model’s training abilities to recognize test images and perform pattern recognition. This feature is not as readily available on the Google platform. Apple Create ML Image classifier’s default training set consists of 75% of total imported images with 5% of the total images being randomly used as a validation set. The remaining 20% of images comprise the testing set. The module’s computational analysis to train the model is achieved in about 2 minutes on average. The score threshold is set at 50% and cannot be manipulated for each image class as in Google AutoML Vision.

Google AutoML Vision is open and can be accessed from many devices. It stores images on remote Google servers but requires computing fees after a $300 credit for 12 months. On AutoML Vision, random 80% of the total images are used in the training set, 10% are used in the validation set, and 10% are used in the testing set. It is important to highlight the different percentages used in the default settings on the respective modules. The time to train the Google AutoML Vision with default computational power is longer on average than Apple Create ML, with about 8 minutes required to train the machine learning module. However, it is possible to choose more computational power for an additional fee and decrease module training time. The user will receive e-mail alerts when the computer time begins and is completed. The computation time is calculated by subtracting the time of the initial e-mail from the final e-mail.

Based on our calculations, we determined there was no significant difference between the 2 machine learning algorithms tested at the default settings with recall and precision values obtained. These findings demonstrate the promise of using a ML algorithm to assist in the performance of human tasks and behaviors, specifically the diagnosis of histopathologic images. These results have numerous potential uses in clinical medicine. ML algorithms have been successfully applied to diagnostic and prognostic endeavors in pathology,^23-28 dermatology,^29-31 ophthalmology,³² cardiology,³³ and radiology.^34-36

Pathologists often use additional tests, such as special staining of tissues or molecular tests, to assist with accurate classification of tumors. ML platforms offer the potential of an additional tool for pathologists to use along with human microscopic interpretation.^37,38 In addition, the number of pathologists in the US is dramatically decreasing, and many other countries have marked physician shortages, especially in fields of specialized training such as pathology.^39-42 These models could readily assist physicians in underserved countries and impact shortages of pathologists elsewhere by providing more specific diagnoses in an expedited manner.⁴³

Finally, although we have explored the application of these platforms in common cancer scenarios, great potential exists to use similar techniques in the detection of other conditions. These include the potential for classification and risk assessment of precancerous lesions, infectious processes in tissue (eg, detection of tuberculosis or malaria),^24,44 inflammatory conditions (eg, arthritis subtypes, gout),⁴⁵ blood disorders (eg, abnormal blood cell morphology),⁴⁶ and many others. The potential of these technologies to improve health care delivery to veteran patients seems to be limited only by the imagination of the user.⁴⁷

Regarding the limited effectiveness in determining the presence or absence of mutations in KRAS for colon adenocarcinoma, it is mentioned that currently pathologists rely on complex molecular tests to detect the mutations at the DNA level.²¹ It is possible that the use of more extensive training data sets may improve recall and precision in cases such as these and warrants further study. Our experiments were limited to the stipulations placed by the free trial software agreements; no costs were expended to use the algorithms, though an Apple computer was required.

Conclusion

We have demonstrated the successful application of 2 readily available ML platforms in providing diagnostic guidance in differentiation between common cancer conditions in veteran patient populations. Although both platforms performed very well with no statistically significant differences in results, some distinctions are worth noting. Apple Create ML can be used on local computers but is limited to an Apple operating system. Google AutoML is not platform-specific but runs only via Google Cloud with associated computational fees. Using these readily available models, we demonstrated the vast potential of AI in diagnostic pathology. The application of AI to clinical medicine remains in the very early stages. The VA is uniquely poised to provide leadership as AI technologies will continue to dramatically change the future of health care, both in veteran and nonveteran patients nationwide.

Acknowledgments

The authors thank Paul Borkowski for his constructive criticism and proofreading of this manuscript. This material is the result of work supported with resources and the use of facilities at the James A. Haley Veterans’ Hospital.

Use of clip closure significantly reduced delayed bleeding in patients who underwent resections for large colorectal lesions, based on data from 235 individuals.

Source: American Gastroenterological Association

“Closure of a mucosal defect with clips after resection has long been considered to reduce the risk of bleeding,” but evidence to support this practice is limited, wrote Eduardo Albéniz, MD, of the Public University of Navarra (Spain), and colleagues.

In a study published in Gastroenterology, the researchers identified 235 consecutive patients who had resections of large nonpedunculated colorectal lesions from May 2016 to June 2018. Patients had an average or high risk of delayed bleeding and were randomized to receive scar closure with either 11-mm through-the-scope clips (119 patients) or no clip (116 patients).

Delayed bleeding occurred in 14 control patients (12.1%), compared with 6 clip patients (5%), for a risk reduction of 7%. The clip group included 68 cases (57%) of complete closure and 33 cases (28%) with partial closure, as well as 18 cases of failure to close (15%); only 1 case of delayed bleeding occurred in the clip group after completion of clip closure. On average, six clips were needed for complete closure.

None of the patients who experienced delayed bleeding required surgical or angiographic intervention, although 15 of the 20 patients with bleeding underwent additional endoscopy. Other adverse events included immediate bleeding in 21 clip patients and 18 controls that was managed with snare soft-tip coagulation. No deaths were reported in connection with the study.

Demographics were similar between the two groups, but the subset of patients with complete closure included more individuals aged 75 years and older and more cases with smaller polyps, compared with other subgroups, the researchers noted.

The study findings were limited by several factors, including the difficulty in predicting delayed bleeding, the potential for selection bias given the timing of patient randomization, the lack of information about polyps that were excluded from treatment, and the difficulty in completely closing the mucosal defects, the researchers noted. However, the results suggest that complete clip closure, despite its challenges, “displays a clear trend to reduce delayed bleeding risk,” and is worth an attempt.

The study was supported by the Spanish Society of Digestive Endoscopy. The researchers had no financial conflicts to disclose. MicroTech (Nanjing, China) contributed the clips used in the study.

SOURCE: Albéniz E et al. Gastroenterology. 2019 Jul 27. doi: 10.1053/j.gastro.2019.07.037.

Use of clip closure significantly reduced delayed bleeding in patients who underwent resections for large colorectal lesions, based on data from 235 individuals.

Source: American Gastroenterological Association

“Closure of a mucosal defect with clips after resection has long been considered to reduce the risk of bleeding,” but evidence to support this practice is limited, wrote Eduardo Albéniz, MD, of the Public University of Navarra (Spain), and colleagues.

In a study published in Gastroenterology, the researchers identified 235 consecutive patients who had resections of large nonpedunculated colorectal lesions from May 2016 to June 2018. Patients had an average or high risk of delayed bleeding and were randomized to receive scar closure with either 11-mm through-the-scope clips (119 patients) or no clip (116 patients).

Delayed bleeding occurred in 14 control patients (12.1%), compared with 6 clip patients (5%), for a risk reduction of 7%. The clip group included 68 cases (57%) of complete closure and 33 cases (28%) with partial closure, as well as 18 cases of failure to close (15%); only 1 case of delayed bleeding occurred in the clip group after completion of clip closure. On average, six clips were needed for complete closure.

None of the patients who experienced delayed bleeding required surgical or angiographic intervention, although 15 of the 20 patients with bleeding underwent additional endoscopy. Other adverse events included immediate bleeding in 21 clip patients and 18 controls that was managed with snare soft-tip coagulation. No deaths were reported in connection with the study.

Demographics were similar between the two groups, but the subset of patients with complete closure included more individuals aged 75 years and older and more cases with smaller polyps, compared with other subgroups, the researchers noted.

The study findings were limited by several factors, including the difficulty in predicting delayed bleeding, the potential for selection bias given the timing of patient randomization, the lack of information about polyps that were excluded from treatment, and the difficulty in completely closing the mucosal defects, the researchers noted. However, the results suggest that complete clip closure, despite its challenges, “displays a clear trend to reduce delayed bleeding risk,” and is worth an attempt.

The study was supported by the Spanish Society of Digestive Endoscopy. The researchers had no financial conflicts to disclose. MicroTech (Nanjing, China) contributed the clips used in the study.

SOURCE: Albéniz E et al. Gastroenterology. 2019 Jul 27. doi: 10.1053/j.gastro.2019.07.037.

Use of clip closure significantly reduced delayed bleeding in patients who underwent resections for large colorectal lesions, based on data from 235 individuals.

Source: American Gastroenterological Association

“Closure of a mucosal defect with clips after resection has long been considered to reduce the risk of bleeding,” but evidence to support this practice is limited, wrote Eduardo Albéniz, MD, of the Public University of Navarra (Spain), and colleagues.

In a study published in Gastroenterology, the researchers identified 235 consecutive patients who had resections of large nonpedunculated colorectal lesions from May 2016 to June 2018. Patients had an average or high risk of delayed bleeding and were randomized to receive scar closure with either 11-mm through-the-scope clips (119 patients) or no clip (116 patients).

Delayed bleeding occurred in 14 control patients (12.1%), compared with 6 clip patients (5%), for a risk reduction of 7%. The clip group included 68 cases (57%) of complete closure and 33 cases (28%) with partial closure, as well as 18 cases of failure to close (15%); only 1 case of delayed bleeding occurred in the clip group after completion of clip closure. On average, six clips were needed for complete closure.

None of the patients who experienced delayed bleeding required surgical or angiographic intervention, although 15 of the 20 patients with bleeding underwent additional endoscopy. Other adverse events included immediate bleeding in 21 clip patients and 18 controls that was managed with snare soft-tip coagulation. No deaths were reported in connection with the study.

Demographics were similar between the two groups, but the subset of patients with complete closure included more individuals aged 75 years and older and more cases with smaller polyps, compared with other subgroups, the researchers noted.

The study findings were limited by several factors, including the difficulty in predicting delayed bleeding, the potential for selection bias given the timing of patient randomization, the lack of information about polyps that were excluded from treatment, and the difficulty in completely closing the mucosal defects, the researchers noted. However, the results suggest that complete clip closure, despite its challenges, “displays a clear trend to reduce delayed bleeding risk,” and is worth an attempt.

The study was supported by the Spanish Society of Digestive Endoscopy. The researchers had no financial conflicts to disclose. MicroTech (Nanjing, China) contributed the clips used in the study.

SOURCE: Albéniz E et al. Gastroenterology. 2019 Jul 27. doi: 10.1053/j.gastro.2019.07.037.

MINNEAPOLIS -- A new era of systemic treatment for bladder cancer has arrived, a US Department of Veterans Affairs (VA) hematologist/oncologist told colleagues, and more changes await on the horizon.

“After a historically long dry spell, you're seeing novel drugs and combinations under investigation,” said Elizabeth Henry, MD, of Edward Hines, Jr. VA Hospital and Loyola University Medical Center in Chicago, Illinois, in a presentation at the annual meeting of the Association of VA Hematology/Oncology. “Our treatment paradigm will almost certainly continue to change.”

There’s a major need for new approaches in bladder cancer, Dr. Henry said. While the median survival for patients with metastatic disease treated has risen, it remains low at about 15 months. And, she said, the 5-year survival rate is about 15% with modern treatments.

Platinum-based chemotherapy is still the first-line treatment, she said, and cisplatin-based combos remain the standard. However, many patients are not eligible to take cisplatin because of factors such as reduced performance status, impaired renal function, peripheral neuropathy, hearing loss and heart failure. “Many patients have renal insufficiency and are platinum ineligible from the get-go,” Dr. Henry said.

In these patients, immune checkpoint inhibitors are an option, but research suggests they may lead to poorer survival in those with PDL1-low tumors. In 2018, the US Food and Drug Administration (FDA) advised their use as first-line treatment only in PDL1-high, cisplatin-ineligible patients or those who can’t undergo chemotherapy, she said.

As second-line therapy after platinum treatment, Dr. Henry said, several drugs targeting the PD1/PDL1 pathway are now FDA-approved with response rates at 15% to 25%; only 1 (pembrolizumab) has level 1 evidence from a phase 3 randomized clinical trial.

Single-agent chemotherapy is an option for patients who have worsened or cannot undergo treatment with immune checkpoint inhibitors. However, according to Dr. Henry, response rates are low (about 10%-15%) and there's no prospective or randomized control trial data showing a survival benefit.

What now? Targeted approaches are entering the picture. For example, fibroblast growth factor receptor inhibitors, which target a pathway that often mutates in bladder cancer. One drug, erdafitinib (Balversa), received FDA approval earlier this year based on a phase 2 trial data that showed an objective response rate (ORR) of 40%. Dr. Henry cautioned that unusual adverse effects can occur, including hyperphosphatemia (a disorder that boosts phosphate levels), ocular toxicity (which can lead to retinal detachment), and toxicity of the skin and hair.

“Patients need to be closely followed if they're starting this as a targeted drug,” Dr. Henry said.

Anti-Nectin-4 antibody drug conjugate, which targets urothelial carcinomas with uniformly high expression of the Nectin-4 cell surface marker, also is showing promise. Recent research suggests a “remarkable” ORR of 42% and nearly 8 months duration of response, she said.

Adverse effects include rash, peripheral neuropathy, and hyperglycemia. “Overall, this is thought to be a relatively well-tolerated therapy,” she said.

In terms of other advances, “we are moving closer to an era of molecular subtype-specific therapeutic strategies,” Dr. Henry said, and the National Comprehensive Cancer Network recommends early molecular testing in stage IIIB/IV urothelial cancer. “It can help facilitate treatment decisions and prevent delays in later lines of therapy, although we're still limited by development of individualized biomarker assays for specific drugs.”

Moving forward, she said, “continued research is needed to learn how to incorporate predictive molecular profiles to optimize treatment selection.”

Dr. Henry reported no relevant disclosures.

MINNEAPOLIS -- A new era of systemic treatment for bladder cancer has arrived, a US Department of Veterans Affairs (VA) hematologist/oncologist told colleagues, and more changes await on the horizon.

“After a historically long dry spell, you're seeing novel drugs and combinations under investigation,” said Elizabeth Henry, MD, of Edward Hines, Jr. VA Hospital and Loyola University Medical Center in Chicago, Illinois, in a presentation at the annual meeting of the Association of VA Hematology/Oncology. “Our treatment paradigm will almost certainly continue to change.”

There’s a major need for new approaches in bladder cancer, Dr. Henry said. While the median survival for patients with metastatic disease treated has risen, it remains low at about 15 months. And, she said, the 5-year survival rate is about 15% with modern treatments.

Platinum-based chemotherapy is still the first-line treatment, she said, and cisplatin-based combos remain the standard. However, many patients are not eligible to take cisplatin because of factors such as reduced performance status, impaired renal function, peripheral neuropathy, hearing loss and heart failure. “Many patients have renal insufficiency and are platinum ineligible from the get-go,” Dr. Henry said.

In these patients, immune checkpoint inhibitors are an option, but research suggests they may lead to poorer survival in those with PDL1-low tumors. In 2018, the US Food and Drug Administration (FDA) advised their use as first-line treatment only in PDL1-high, cisplatin-ineligible patients or those who can’t undergo chemotherapy, she said.

As second-line therapy after platinum treatment, Dr. Henry said, several drugs targeting the PD1/PDL1 pathway are now FDA-approved with response rates at 15% to 25%; only 1 (pembrolizumab) has level 1 evidence from a phase 3 randomized clinical trial.

Single-agent chemotherapy is an option for patients who have worsened or cannot undergo treatment with immune checkpoint inhibitors. However, according to Dr. Henry, response rates are low (about 10%-15%) and there's no prospective or randomized control trial data showing a survival benefit.

What now? Targeted approaches are entering the picture. For example, fibroblast growth factor receptor inhibitors, which target a pathway that often mutates in bladder cancer. One drug, erdafitinib (Balversa), received FDA approval earlier this year based on a phase 2 trial data that showed an objective response rate (ORR) of 40%. Dr. Henry cautioned that unusual adverse effects can occur, including hyperphosphatemia (a disorder that boosts phosphate levels), ocular toxicity (which can lead to retinal detachment), and toxicity of the skin and hair.

“Patients need to be closely followed if they're starting this as a targeted drug,” Dr. Henry said.

Anti-Nectin-4 antibody drug conjugate, which targets urothelial carcinomas with uniformly high expression of the Nectin-4 cell surface marker, also is showing promise. Recent research suggests a “remarkable” ORR of 42% and nearly 8 months duration of response, she said.

Adverse effects include rash, peripheral neuropathy, and hyperglycemia. “Overall, this is thought to be a relatively well-tolerated therapy,” she said.

In terms of other advances, “we are moving closer to an era of molecular subtype-specific therapeutic strategies,” Dr. Henry said, and the National Comprehensive Cancer Network recommends early molecular testing in stage IIIB/IV urothelial cancer. “It can help facilitate treatment decisions and prevent delays in later lines of therapy, although we're still limited by development of individualized biomarker assays for specific drugs.”

Moving forward, she said, “continued research is needed to learn how to incorporate predictive molecular profiles to optimize treatment selection.”

Dr. Henry reported no relevant disclosures.

MINNEAPOLIS -- A new era of systemic treatment for bladder cancer has arrived, a US Department of Veterans Affairs (VA) hematologist/oncologist told colleagues, and more changes await on the horizon.

“After a historically long dry spell, you're seeing novel drugs and combinations under investigation,” said Elizabeth Henry, MD, of Edward Hines, Jr. VA Hospital and Loyola University Medical Center in Chicago, Illinois, in a presentation at the annual meeting of the Association of VA Hematology/Oncology. “Our treatment paradigm will almost certainly continue to change.”

There’s a major need for new approaches in bladder cancer, Dr. Henry said. While the median survival for patients with metastatic disease treated has risen, it remains low at about 15 months. And, she said, the 5-year survival rate is about 15% with modern treatments.

Platinum-based chemotherapy is still the first-line treatment, she said, and cisplatin-based combos remain the standard. However, many patients are not eligible to take cisplatin because of factors such as reduced performance status, impaired renal function, peripheral neuropathy, hearing loss and heart failure. “Many patients have renal insufficiency and are platinum ineligible from the get-go,” Dr. Henry said.

In these patients, immune checkpoint inhibitors are an option, but research suggests they may lead to poorer survival in those with PDL1-low tumors. In 2018, the US Food and Drug Administration (FDA) advised their use as first-line treatment only in PDL1-high, cisplatin-ineligible patients or those who can’t undergo chemotherapy, she said.

As second-line therapy after platinum treatment, Dr. Henry said, several drugs targeting the PD1/PDL1 pathway are now FDA-approved with response rates at 15% to 25%; only 1 (pembrolizumab) has level 1 evidence from a phase 3 randomized clinical trial.

Single-agent chemotherapy is an option for patients who have worsened or cannot undergo treatment with immune checkpoint inhibitors. However, according to Dr. Henry, response rates are low (about 10%-15%) and there's no prospective or randomized control trial data showing a survival benefit.

What now? Targeted approaches are entering the picture. For example, fibroblast growth factor receptor inhibitors, which target a pathway that often mutates in bladder cancer. One drug, erdafitinib (Balversa), received FDA approval earlier this year based on a phase 2 trial data that showed an objective response rate (ORR) of 40%. Dr. Henry cautioned that unusual adverse effects can occur, including hyperphosphatemia (a disorder that boosts phosphate levels), ocular toxicity (which can lead to retinal detachment), and toxicity of the skin and hair.

“Patients need to be closely followed if they're starting this as a targeted drug,” Dr. Henry said.

Anti-Nectin-4 antibody drug conjugate, which targets urothelial carcinomas with uniformly high expression of the Nectin-4 cell surface marker, also is showing promise. Recent research suggests a “remarkable” ORR of 42% and nearly 8 months duration of response, she said.

Adverse effects include rash, peripheral neuropathy, and hyperglycemia. “Overall, this is thought to be a relatively well-tolerated therapy,” she said.

In terms of other advances, “we are moving closer to an era of molecular subtype-specific therapeutic strategies,” Dr. Henry said, and the National Comprehensive Cancer Network recommends early molecular testing in stage IIIB/IV urothelial cancer. “It can help facilitate treatment decisions and prevent delays in later lines of therapy, although we're still limited by development of individualized biomarker assays for specific drugs.”

Moving forward, she said, “continued research is needed to learn how to incorporate predictive molecular profiles to optimize treatment selection.”

Dr. Henry reported no relevant disclosures.

Screening for colorectal cancer in patients with cystic fibrosis is cost effective, and should be started at a younger age and performed more often, new research suggests.

While colorectal cancer (CRC) screening traditionally begins at age 50 years in people at average risk for the disease, those at high risk usually begin undergoing colonoscopies at an earlier age. Patients with cystic fibrosis fall under the latter category, wrote Andrea Gini, of the department of public health at Erasmus Medical Center in Rotterdam, the Netherlands, and colleagues, with an incidence of CRC up to 30 times higher than the general population, but their shorter lifespan has led to a “different trade-off between the benefits and harms of CRC screening.”

Between 2000 and 2015, the median predicted survival age for patients with cystic fibrosis increased from 33.3 years to 41.7 years; this increased survival has brought increased risk for other diseases, particularly in the GI tract, Mr. Gini and colleagues wrote in Gastroenterology. By using the Microsimulation Screening Analysis–Colon model – a joint project between Erasmus Medical Center and Memorial Sloan Kettering Cancer Center in New York – the investigators assessed the cost-effectiveness of CRC screening in patients with cystic fibrosis.

Three cohorts of 10 million patients each were simulated, with one cohort having undergone transplant, one cohort not having transplant, and one cohort of individuals without cystic fibrosis. The simulated patient age was 30 years in 2017. A total of 76 different colonoscopy-screening strategies were assessed, with each differing in screening interval (3, 5, or 10 years for colonoscopy), age to start screening (30, 35, 40, 45, or 50 years), and age to end screening (55, 60, 65, 70, or 75 years). The optimal screening strategy was determined based on a willingness-to-pay threshold of $100,000 per life-year gained, the investigators wrote.

In the absence of screening, the mortality rate for nontransplant cystic fibrosis patients was 19.1 per 1,000 people, and the rate for cystic fibrosis patients who had undergone transplant was 22.3 per 1,000 people. The standard screening strategy prevented more than 73% of CRC deaths in the general population, 66% of deaths in nontransplant cystic fibrosis patients, and 36% of deaths in cystic fibrosis patients with transplant; however, the model predicted that only 22% of individuals who received a transplant and 36% of those who did not would reach the age of 50 years.

According to the model, the optimal colonoscopy-screening strategy for nontransplant patients was one screen every 5 years, starting at 40 and screening until the age of 75. The incremental cost-effectiveness ratio (ICER) was $84,000 per life-year gained; CRC incidence was reduced by 52% and CRC mortality was reduced by 79%. For transplant patients, the best strategy was one screen every 3 years between the ages of 35 and 55, which reduced CRC mortality by 82% at an ICER of $71,000 per life-year gained.

In a separate analysis of fecal immunochemical testing, a less-demanding alternative to colonoscopy, the optimal screening strategy was an annual test between the age of 35 and 75 years for nontransplant cystic fibrosis patients, for an ICER of $47,000 per life-year gained and a CRC mortality reduction of 78%. The best strategy for transplant patients was once a year between the ages of 30 and 60, which reduced CRC mortality by 77% at an ICER of $86,000 per life-year gained. While fecal immunochemical testing may be more cost effective than colonoscopy, “specific evidence of its performance in the cystic fibrosis population is required before considering this screening modality,” the investigators noted.

“This study indicates that there is benefit to earlier CRC screening in the cystic fibrosis population and [that it] can be done at acceptable costs,” the investigators wrote. “The findings of this analysis support clinicians, researchers, and policy makers who aim to define a tailored CRC screening for individuals with cystic fibrosis in the United States.”

The study was funded by the Cystic Fibrosis Foundation, the Cancer Intervention and Surveillance Modeling Network consortium, and Memorial Sloan Kettering Cancer Center. The investigators reported no conflicts of interest.

[email protected]

SOURCE: Gini A et al. Gastroenterology. 2017 Dec 27. doi: 10.1053/j.gastro.2017.12.011.

Screening for colorectal cancer in patients with cystic fibrosis is cost effective, and should be started at a younger age and performed more often, new research suggests.

While colorectal cancer (CRC) screening traditionally begins at age 50 years in people at average risk for the disease, those at high risk usually begin undergoing colonoscopies at an earlier age. Patients with cystic fibrosis fall under the latter category, wrote Andrea Gini, of the department of public health at Erasmus Medical Center in Rotterdam, the Netherlands, and colleagues, with an incidence of CRC up to 30 times higher than the general population, but their shorter lifespan has led to a “different trade-off between the benefits and harms of CRC screening.”

Between 2000 and 2015, the median predicted survival age for patients with cystic fibrosis increased from 33.3 years to 41.7 years; this increased survival has brought increased risk for other diseases, particularly in the GI tract, Mr. Gini and colleagues wrote in Gastroenterology. By using the Microsimulation Screening Analysis–Colon model – a joint project between Erasmus Medical Center and Memorial Sloan Kettering Cancer Center in New York – the investigators assessed the cost-effectiveness of CRC screening in patients with cystic fibrosis.

Three cohorts of 10 million patients each were simulated, with one cohort having undergone transplant, one cohort not having transplant, and one cohort of individuals without cystic fibrosis. The simulated patient age was 30 years in 2017. A total of 76 different colonoscopy-screening strategies were assessed, with each differing in screening interval (3, 5, or 10 years for colonoscopy), age to start screening (30, 35, 40, 45, or 50 years), and age to end screening (55, 60, 65, 70, or 75 years). The optimal screening strategy was determined based on a willingness-to-pay threshold of $100,000 per life-year gained, the investigators wrote.

In the absence of screening, the mortality rate for nontransplant cystic fibrosis patients was 19.1 per 1,000 people, and the rate for cystic fibrosis patients who had undergone transplant was 22.3 per 1,000 people. The standard screening strategy prevented more than 73% of CRC deaths in the general population, 66% of deaths in nontransplant cystic fibrosis patients, and 36% of deaths in cystic fibrosis patients with transplant; however, the model predicted that only 22% of individuals who received a transplant and 36% of those who did not would reach the age of 50 years.

According to the model, the optimal colonoscopy-screening strategy for nontransplant patients was one screen every 5 years, starting at 40 and screening until the age of 75. The incremental cost-effectiveness ratio (ICER) was $84,000 per life-year gained; CRC incidence was reduced by 52% and CRC mortality was reduced by 79%. For transplant patients, the best strategy was one screen every 3 years between the ages of 35 and 55, which reduced CRC mortality by 82% at an ICER of $71,000 per life-year gained.

In a separate analysis of fecal immunochemical testing, a less-demanding alternative to colonoscopy, the optimal screening strategy was an annual test between the age of 35 and 75 years for nontransplant cystic fibrosis patients, for an ICER of $47,000 per life-year gained and a CRC mortality reduction of 78%. The best strategy for transplant patients was once a year between the ages of 30 and 60, which reduced CRC mortality by 77% at an ICER of $86,000 per life-year gained. While fecal immunochemical testing may be more cost effective than colonoscopy, “specific evidence of its performance in the cystic fibrosis population is required before considering this screening modality,” the investigators noted.

“This study indicates that there is benefit to earlier CRC screening in the cystic fibrosis population and [that it] can be done at acceptable costs,” the investigators wrote. “The findings of this analysis support clinicians, researchers, and policy makers who aim to define a tailored CRC screening for individuals with cystic fibrosis in the United States.”

The study was funded by the Cystic Fibrosis Foundation, the Cancer Intervention and Surveillance Modeling Network consortium, and Memorial Sloan Kettering Cancer Center. The investigators reported no conflicts of interest.

[email protected]

SOURCE: Gini A et al. Gastroenterology. 2017 Dec 27. doi: 10.1053/j.gastro.2017.12.011.

Screening for colorectal cancer in patients with cystic fibrosis is cost effective, and should be started at a younger age and performed more often, new research suggests.

While colorectal cancer (CRC) screening traditionally begins at age 50 years in people at average risk for the disease, those at high risk usually begin undergoing colonoscopies at an earlier age. Patients with cystic fibrosis fall under the latter category, wrote Andrea Gini, of the department of public health at Erasmus Medical Center in Rotterdam, the Netherlands, and colleagues, with an incidence of CRC up to 30 times higher than the general population, but their shorter lifespan has led to a “different trade-off between the benefits and harms of CRC screening.”

Between 2000 and 2015, the median predicted survival age for patients with cystic fibrosis increased from 33.3 years to 41.7 years; this increased survival has brought increased risk for other diseases, particularly in the GI tract, Mr. Gini and colleagues wrote in Gastroenterology. By using the Microsimulation Screening Analysis–Colon model – a joint project between Erasmus Medical Center and Memorial Sloan Kettering Cancer Center in New York – the investigators assessed the cost-effectiveness of CRC screening in patients with cystic fibrosis.

Three cohorts of 10 million patients each were simulated, with one cohort having undergone transplant, one cohort not having transplant, and one cohort of individuals without cystic fibrosis. The simulated patient age was 30 years in 2017. A total of 76 different colonoscopy-screening strategies were assessed, with each differing in screening interval (3, 5, or 10 years for colonoscopy), age to start screening (30, 35, 40, 45, or 50 years), and age to end screening (55, 60, 65, 70, or 75 years). The optimal screening strategy was determined based on a willingness-to-pay threshold of $100,000 per life-year gained, the investigators wrote.

In the absence of screening, the mortality rate for nontransplant cystic fibrosis patients was 19.1 per 1,000 people, and the rate for cystic fibrosis patients who had undergone transplant was 22.3 per 1,000 people. The standard screening strategy prevented more than 73% of CRC deaths in the general population, 66% of deaths in nontransplant cystic fibrosis patients, and 36% of deaths in cystic fibrosis patients with transplant; however, the model predicted that only 22% of individuals who received a transplant and 36% of those who did not would reach the age of 50 years.

According to the model, the optimal colonoscopy-screening strategy for nontransplant patients was one screen every 5 years, starting at 40 and screening until the age of 75. The incremental cost-effectiveness ratio (ICER) was $84,000 per life-year gained; CRC incidence was reduced by 52% and CRC mortality was reduced by 79%. For transplant patients, the best strategy was one screen every 3 years between the ages of 35 and 55, which reduced CRC mortality by 82% at an ICER of $71,000 per life-year gained.

In a separate analysis of fecal immunochemical testing, a less-demanding alternative to colonoscopy, the optimal screening strategy was an annual test between the age of 35 and 75 years for nontransplant cystic fibrosis patients, for an ICER of $47,000 per life-year gained and a CRC mortality reduction of 78%. The best strategy for transplant patients was once a year between the ages of 30 and 60, which reduced CRC mortality by 77% at an ICER of $86,000 per life-year gained. While fecal immunochemical testing may be more cost effective than colonoscopy, “specific evidence of its performance in the cystic fibrosis population is required before considering this screening modality,” the investigators noted.

“This study indicates that there is benefit to earlier CRC screening in the cystic fibrosis population and [that it] can be done at acceptable costs,” the investigators wrote. “The findings of this analysis support clinicians, researchers, and policy makers who aim to define a tailored CRC screening for individuals with cystic fibrosis in the United States.”

The study was funded by the Cystic Fibrosis Foundation, the Cancer Intervention and Surveillance Modeling Network consortium, and Memorial Sloan Kettering Cancer Center. The investigators reported no conflicts of interest.

[email protected]

SOURCE: Gini A et al. Gastroenterology. 2017 Dec 27. doi: 10.1053/j.gastro.2017.12.011.