On 2014 Apr 04, David Keller commented:

Move Leftward on MultiTarget's Receiver Operating Curve

The MultiTarget test has higher sensitivity but lower specificity than FIT, the standard colon cancer fecal screen (1). MultiTarget detects more colon malignancies but triggers more unnecessary colonoscopies. Colonoscopy is expensive and invasive, limiting the acceptability of false-positive fecal screens. MultiTarget’s Composite Score threshold was set at 183. Increasing this threshold score will increase specificity and lower sensitivity (2). It should be raised until MultiTarget’s specificity equals FIT’s; if, at that point, MultiTarget’s sensitivity remains higher than FIT’s, then replacing FIT screening with MultiTarget will detect more malignancies without forcing patients to undergo more unnecessary colonoscopies.

MultiTarget incorporates fecal hemoglobin testing. Reducing the cofactor X6 in the Logistic Score has the effect of increasing specificity and reducing sensitivity of the fecal hemoglobin component of MultiTarget without affecting the performance of the DNA components of the Composite Score.

Fecal hemoglobin testing should be eliminated from colon cancer stool screening protocols as DNA tests improve, because bleeding commonly originates from benign sources, such as hemorrhoids, degrading specificity.

References

1) Imperiale TF, Ransohoff DF, Itzkowitz SH, Levin TR, Lavin P, Lidgard GP, Ahlquist DA, Berger BM. Multitarget stool DNA testing for colorectal-cancer screening. N Engl J Med. 2014 Apr 3;370(14):1287-97. doi: 10.1056/NEJMoa1311194. Epub 2014 Mar 19. PubMed PMID: 24645800.

2) Florkowski CM. Sensitivity, specificity, receiver-operating characteristic (ROC) curves and likelihood ratios: communicating the performance of diagnostic tests. Clin Biochem Rev. 2008 Aug;29 Suppl 1:S83-7. PubMed PMID: 18852864; PubMed Central PMCID: PMC2556590.

On 2014 Apr 03, David Keller commented:

On 2014 Apr 02, David Keller commented:

We need a colon cancer stool screen which does not equate fecal blood with malignancy

The new "multitarget stool DNA test" studied in this clinical trial actually includes a hemoglobin immunoassay, so it retains a degree of the inherent inaccuracy caused by equating the presence of fecal blood with the presence of cancer. This new screening test for colon cancer uses a composite score computed from the presence of direct molecular markers of cancer (such as cancer-associated DNA strands), but the score also includes a component from the presence of fecal blood. The fecal blood test component cannot distinguish between blood from a benign source, such as a hemorrhoid, versus bleeding from a malignancy; this tends to increase the false-positive rate of the multitarget test, which was actually worse than that of the standard FIT test used as for comparison.

Hemorrhoids are common, and their trace bleeding increases the number of benign (unnecessary) colonoscopies which are ordered as a result of screening for fecal hemoglobin. The presence of stool DNA specific for colon cancer will be a convincing argument in favor of colonoscopy for patients who attribute the presence of blood in their stool to their hemorrhoids. The ideal stool screening test would rely purely on detecting cancer-specific molecules, without the need to enhance sensitivity by including a test for fecal blood.

A Modest Proposal - As a clinician, I would want to see the results of all the individual components of the multitarget stool screening test reported separately, along with the composite score calculated using the complex formula employed in this study. The investigators evidently fine-tuned this formula by altering the constant coefficients applied to each test component, in order to achieve what they hoped would be an optimized combination of sensitivity and specificity. However, whenever several directly-measured quantities are combined into a single number for the purpose of arriving at a binary result, much information is lost. The binary result will be a useful "yes or no" answer to the question of whether a patient requires colonoscopy. But clinicians should also be informed of the individual results of each of the component tests, along with the characteristics and implications of each, to allow a more individualized approach to the care of certain patients.

On 2014 Apr 02, David Keller commented:

We need a colon cancer stool screen which does not equate fecal blood with malignancy

The new "multitarget stool DNA test" studied in this clinical trial actually includes a hemoglobin immunoassay, so it retains a degree of the inherent inaccuracy caused by equating the presence of fecal blood with the presence of cancer. This new screening test for colon cancer uses a composite score computed from the presence of direct molecular markers of cancer (such as cancer-associated DNA strands), but the score also includes a component from the presence of fecal blood. The fecal blood test component cannot distinguish between blood from a benign source, such as a hemorrhoid, versus bleeding from a malignancy; this tends to increase the false-positive rate of the multitarget test, which was actually worse than that of the standard FIT test used as for comparison.

Hemorrhoids are common, and their trace bleeding increases the number of benign (unnecessary) colonoscopies which are ordered as a result of screening for fecal hemoglobin. The presence of stool DNA specific for colon cancer will be a convincing argument in favor of colonoscopy for patients who attribute the presence of blood in their stool to their hemorrhoids. The ideal stool screening test would rely purely on detecting cancer-specific molecules, without the need to enhance sensitivity by including a test for fecal blood.

A Modest Proposal - As a clinician, I would want to see the results of all the individual components of the multitarget stool screening test reported separately, along with the composite score calculated using the complex formula employed in this study. The investigators evidently fine-tuned this formula by altering the constant coefficients applied to each test component, in order to achieve what they hoped would be an optimized combination of sensitivity and specificity. However, whenever several directly-measured quantities are combined into a single number for the purpose of arriving at a binary result, much information is lost. The binary result will be a useful "yes or no" answer to the question of whether a patient requires colonoscopy. But clinicians should also be informed of the individual results of each of the component tests, along with the characteristics and implications of each, to allow a more individualized approach to the care of certain patients.

On 2014 Apr 03, David Keller commented:

On 2014 Apr 04, David Keller commented:

Move Leftward on MultiTarget's Receiver Operating Curve

The MultiTarget test has higher sensitivity but lower specificity than FIT, the standard colon cancer fecal screen (1). MultiTarget detects more colon malignancies but triggers more unnecessary colonoscopies. Colonoscopy is expensive and invasive, limiting the acceptability of false-positive fecal screens. MultiTarget’s Composite Score threshold was set at 183. Increasing this threshold score will increase specificity and lower sensitivity (2). It should be raised until MultiTarget’s specificity equals FIT’s; if, at that point, MultiTarget’s sensitivity remains higher than FIT’s, then replacing FIT screening with MultiTarget will detect more malignancies without forcing patients to undergo more unnecessary colonoscopies.

MultiTarget incorporates fecal hemoglobin testing. Reducing the cofactor X6 in the Logistic Score has the effect of increasing specificity and reducing sensitivity of the fecal hemoglobin component of MultiTarget without affecting the performance of the DNA components of the Composite Score.

Fecal hemoglobin testing should be eliminated from colon cancer stool screening protocols as DNA tests improve, because bleeding commonly originates from benign sources, such as hemorrhoids, degrading specificity.

References

1) Imperiale TF, Ransohoff DF, Itzkowitz SH, Levin TR, Lavin P, Lidgard GP, Ahlquist DA, Berger BM. Multitarget stool DNA testing for colorectal-cancer screening. N Engl J Med. 2014 Apr 3;370(14):1287-97. doi: 10.1056/NEJMoa1311194. Epub 2014 Mar 19. PubMed PMID: 24645800.

2) Florkowski CM. Sensitivity, specificity, receiver-operating characteristic (ROC) curves and likelihood ratios: communicating the performance of diagnostic tests. Clin Biochem Rev. 2008 Aug;29 Suppl 1:S83-7. PubMed PMID: 18852864; PubMed Central PMCID: PMC2556590.