Authored by Dr. Mark Noar, Inventor of ENDOSURE
Perhaps one of the most common questions that we face, when clinicians first experience the near 100% accuracy of the Endosure test, is, why is there a pain score? Does this pain score not introduce subjectivity into an objective test?
This question to be quite curious and interesting, because in the field of endometriosis clinical decisions rely heavily upon the presence of pain to determine if surgery is indicated or treatment administered. With very rare exceptions, it would be nearly impossible to find a surgeon who will operate on a woman with endometriosis who does not have some complaints of pain. Even decisions to operate to correct problems of fertility according to the newest society guidelines require the presence of pain in the absence of structural abnormalities seen on other imaging studies. Why is the subjective nature of pain not equally problematic in the decision to perform surgery?
First, it’s important to understand that the pain score used in the Endosure test predictive analysis is based on the ENDO4D validated pain questionnaire using a visual pain scale. This pain questionnaire has long been accepted and validated in the literature.
Second, is to understand how the pain score became incorporated into the Endosure’s complex predictive calculation. To do that, it’s necessary to understand the development of the test. In a 1998 paper published in Fertility and Sterility, using transnasal motility catheters over a 24-hour period of time, it was noticed that women with endometriosis had a particular smooth muscle abnormality of contraction of the small bowel. This abnormality was not found in any other concomitant disease. The origin of this abnormal and highly specific and irregular, small bowel frequency was determined to be due to the simultaneous secretion of large quantities of two prostaglandins by the endometriosis tissue. The first prostaglandin is F alpha which causes simultaneous contraction of both the circular as well as the longitudinal smooth muscle of the small bowel. This type of contractility is very similar to that caused by dysentery. The second is prostaglandin E2 which is responsible for the creation of normal peristalsis. When both are secreted in high quantities, there is simultaneous peristalsis and spasm that takes place leading to a standing wave pattern.
FIGURE 1
This is the abnormal waveform that was documented by transnasal motility catheter. It was possible to use an existing device, the EVG or Electroviscerogram, using abdominal electrodes to sense the same energy known as the gastrointestinal myoelectrical activity (Fig 1).
Once it was established that this was possible and reproducible, a period of 10 years of research began using this technology in over 4000 both men and women with endometriosis as well as other potentially confounding diseases. Once again it was demonstrated that women with endometriosis had an exclusive frequency pattern between the frequencies of 20 and 60 cycles per minute. These frequencies were not found to occur in any other illness other than adenomyosis, itself a form of endometriosis.
At the end of this period of time a controlled trial was designed and approved by the investigational review board. In this trial there were three groups. There were women without endometriosis, some of whom had other confounding, illnesses, such as IBS or IBD who served as a control group. The second group consisted of women who had just undergone surgical biopsy, demonstrating endometriosis.
They had their EVG study after surgery. The final group was a validation group of women with abdominal pain of uncertain origin with no diagnosis on imaging. This group underwent EVG followed by subsequent surgery. The important point here is that with the exception of the control group, all subjects were surgically verified. Also of note is that the surgeon, as well as those performing the EVG studies were unaware of the results of the other.
Following completion of the 154-patient trial, all of the data was collected, including the age, pain scores, and the percent distribution of the coefficient of the area under the curve of the small bowel contractions. All of this was placed into a formula using an artificial intelligence driven program to search and evaluate combinations of all variables that would allow the development of a predictive model in which all positives tested as positive and all negatives tested as negative (Fig 2).
FIGURE 2
FIGURE 3
From this model, it was determined that the strongest statistical result occurred when the pain score, the age, and the frequencies of the area under the curve at the 30 to 40 and 40 to 50 cpm frequencies were used. The formula was constructed in this matter because it gave the highest accurate predictability. During the calculations, other formulations were used and evaluated that did not include the pain score and or the age, and these were found to be much less accurate (Fig 3).
The conundrum at the time was do you include the pain score which provided for 98 to 99+ percent accuracy or do you exclude the pain score and accept 79% or 88% accuracy? The decision for the current predictive model was to include both age and pain score which gave the highest accuracy with the lowest false positive and false negative rate.
For those who prefer to not include the pain score, it is possible to provide an alternative formula that is considerably less accurate, but still provides a 79%-88% accuracy as noted in the accompanying graphic. Predictive modeling is a dynamic living and breathing tool. As more and more studies are conducted, there is more and more data to improve the score and one day the pain score may become obsolete.
Perhaps the question that each physician will need to answer is do they want to include a pain score calculation that gives them a 99+ percent accurate test, or do they want to exclude the pain score in which case the accuracy will drop to 79-88%? Behind this question is another question, and that is how do you plan on explaining to your patient that you chose to use an inferior test with a higher risk of false positives and false negatives, when a better one is available that incorporates pain scores and is based upon surgical validation?