G. S. Duthie

Artificial neural networks applied to outcome prediction for colorectal cancer patients in separate institutions

BACKGROUND Artificial neural networks are computer programs that can be used to discover complex relations within data sets. They permit the recognition of patterns in complex biological data sets that cannot be detected with conventional linear statistical analysis. One such complex problem is the prediction of outcome for individual patients treated for colorectal cancer. Predictions of outcome in such patients have traditionally been based on population statistics. However, these predictions have little meaning for the individual patient. We report the training of neural networks to predict outcome for individual patients from one institution and their predictive performance on data from a different institution in another region. METHODS 5-year follow-up data from 334 patients treated for colorectal cancer were used to train and validate six neural networks designed for the prediction of death within 9, 12, 15, 18, 21, and 24 months. The previously trained 12-month neural network was then applied to 2-year follow-up data from patients from a second institution; outcome was concealed. No further training of the neural network was undertaken. The networks predictions were compared with those of two consultant colorectal surgeons supplied with the same data. FINDINGS All six neural networks were able to achieve overall accuracy greater than 80% for the prediction of death for individual patients at institution 1 within 9, 12, 15, 18, 21, and 24 months. The mean sensitivity and specificity were 60% and 88%. When the neural network trained to predict death within 12 months was applied to data from the second institution, overall accuracy of 90% (95% CI 84-96) was achieved, compared with the overall accuracy of the colorectal surgeons of 79% (71-87) and 75% (66-84). INTERPRETATION The neural networks were able to predict outcome for individual patients with colorectal cancer much more accurately than the currently available clinicopathological methods. Once trained on data from one institution, the neural networks were able to predict outcome for patients from an unrelated institution.

Obstetric anal sphincter injury

PURPOSE: An obstetrically damaged anal sphincter is the principal cause of the development of fecal incontinence in otherwise healthy females. Reports suggest that such damage complicates as many as 35 percent of primiparous vaginal deliveries, with 13 percent of first-time mothers becoming symptomatic. In maternity units delivering 3,000 patients annually, it would follow that 390 symptomatic patients would develop new symptoms each year. This incidence of dysfunction does not reflect current clinical practice. We have investigated this discrepancy to establish the actual incidence of anal sphincter trauma associated with childbirth. METHODS: During a six-week period, 159 females (105 primiparous and 54 para-I) were prospectively assessed postnatally using a standardized symptom questionnaire, endoanal ultrasound, and anal manometry. This group constituted 84 percent of all eligible deliveries occurring in the unit during the study period. RESULTS: One patient developed fecal urgency after this delivery; there were no reports of fecal incontinence. Anal sphincter injuries were identified ultrasonically in 6.8 percent of primiparous patients, 12.2 percent of para-I patients having vaginal deliveries, and 83 percent of patients having forceps deliveries overall. Manometric data provided confirmatory evidence, with significantly reduced maximum squeeze pressures in patients with a disrupted anal sphincter (P<0.0005). CONCLUSIONS: A symptom questionnaire is inadequate to identify anal sphincter injuries. The incidence of sphincter injury in relation to vaginal delivery has been overestimated in previous published work. This study demonstrates that the true incidence is 8.7 percent overall and that symptoms of sphincter dysfunction are uncommon—this is in keeping with current clinical practice.

Endosonographic evidence of injury to the internal anal sphincter after low anterior resection: long-term follow-up.

PURPOSE: Transanal stapled anastomosis has been associated with continence disturbances and reduced postoperative anal sphincter function. The aim of the present work was to study the effect of transanal stapling on anal sphincter morphology by endoanal ultrasound. METHODS: Thirty-nine consecutive patients undergoing stapled low anterior resection for rectal carcinoma were assessed. Each patient was assessed by endoluminal ultrasound before surgery, immediately after surgery, and at 3, 6, 9, 12, and 24 months after surgery. RESULTS: There were no preoperative internal and sphincter defects observed. Three female patients were observed to have preoperative evidence of external anal sphincter defects. After low anterior resection, seven patients were found to have internal anal sphincter defects, which persisted after the two-year follow-up. There were no additional external anal sphincter injuries. Three patients with internal anal sphincter injuries required the use of pads for poor bowel function. CONCLUSIONS: Up to 18 percent of patients who underwent stapled low anterior resection had long-term evidence of internal anal sphincter injury. The external sphincter does not appear to be affected by the procedure.

Normal female anal sphincter: difficulties in interpretation explained.

AbstractPURPOSE: Our aims were to quantify the nature, characteristics, and frequency of variations in female anal sphincter anatomy. METHODS: Nulliparous patients from the antenatal clinic and healthy volunteers of both genders were studied. Sphincter length was determined by the position of the puborectalis sling. Defects in the external anal sphincter were defined at each level and recorded in degrees. Cylindric longitudinal images of the endoanal scans were created by a three-dimensional-representation software package. Manometry was performed by a pull-through technique. RESULTS: Fifty-seven nulliparous patients and 18 healthy volunteers were included in the study. The mean age was 39 years for males and 28.35 years for females. There was no significant difference in overall sphincter length or in the internal anal sphincter length as a percentage of overall sphincter length between genders. All nine males had a complete ring of external anal sphincter along the full sphincter length. In the external anal sphincter below the level of the puborectalis sling, a natural gap occurred in 43 nulliparous (75 percent) and all 9 female volunteers. The greater the size of the defect, the greater its extent (mean 1.33 cm for >90° and 1.16 cm for <90°; chi-squared P = 0.008, eight degrees of freedom). Manometry provided confirmatory evidence of the gaps seen. Anal manometry was analyzed by Mann-Whitney U test for continuous nonparametric data and t-test for comparison between genders. CONCLUSION: The female sphincter has a variable natural defect occurring along its anterior length. This makes interpretation of the isolated endoanal ultrasound difficult and explains previous overreporting of obstetric sphincter defects.

Preliminary experience with butyl-2-cyanoacrylate adhesive in tension-free inguinal hernia repair

Sir We read with interest the paper by Dr Shifrin et al. (Br J Sue 1996; 83: 1107-9) regarding carotid endarterectomy without preoperative angiography using Duplex ultrasonography as the sole investigative technique. They concluded that it was safe to omit angiography in preoperative assessment (in patients selected on surgeon’s preference) as they found no difference in postoperative complications compared with patients who had also undergone conventional angiography. We do not consider that this conclusion can be drawn from their results. Following the North American Symptomatic Carotid Endarterectomy Trial study’ a clearly defined group of patients with more than 70 per cent stenosis of the internal carotid artery were demonstrated to receive benefit from surgery compared with medical therapy. An important role of preoperative investigation in carotid disease is to assess accurately the degree of stenosis to define those patients who will benefit from surgery. The authors do not state how they assessed the stenosis found at operation and how closely it agreed with the Duplex findings. How do the authors know that they have not operated on a number of patients with a 50 per cent stenosis who did not require surgery? Such a group of patients would be exposed to the risks of surgery unnecessarily. From the data provided in this paper we would therefore still be reticent to perform endarterectomy on patients based on a single preoperative investigation such as duplex ultrasonography.

Does laparoscopic abdominoperineal resection of the rectum compromise long-term survival?

AbstractPURPOSE: Laparoscopic techniques for bowel resection have not enjoyed widespread popularity. Of concern is that long-term follow-up data of cancer specific outcomes is not yet available. The aim of our study was to examine the long-term outcome of abdominoperineal resection for cancer done laparoscopically compared with a similar cohort who underwent open surgery. METHODS: A retrospective review was performed of all abdominoperineal resections done in our center between 1992 and 2000, comparing the cancer-specific outcomes of the laparoscopic cohort with the open cohort. The analysis was performed on an intention-to-treat basis and survival analysis was calculated by the techniques of Kaplan-Meier. RESULTS: Eighty-nine patients were reviewed. Twenty-eight operations were done laparoscopically, and 61 were open. The two groups were matched for age and stage of disease. There was no difference in mean length of overall survival (open = 30.3 months; laparoscopic = 40.8 months; P = 0.355 log rank). No difference in overall recurrence rate, isolated recurrence rate, or distant recurrence rates was seen nor was there any difference in the disease-free period. There was no difference in the number of lymph nodes harvested from the resected specimens, and the distance to the lateral margins or involvement of tumor in the lateral margins between the two groups was the same. The laparoscopic cohort did have a significantly shorter length of stay (mean, 13 days) compared with the open cohort (mean, 18 days), P = 0.008 Mann-Whitney U test. CONCLUSIONS: Laparoscopic abdominoperineal resection of the rectum for cancer does not compromise cancer-specific survival outcomes. The patients avoid a large abdominal wound, which improves cosmesis and presumably is responsible for the significantly shorter length of stay.

Rectoanal reflex parameters in incontinence and constipation.

AbstractPURPOSE: The transient relaxation of the internal anal sphincter in response to rectal distention is believed to play an important role in the continence mechanism. Most anorectal physiology laboratories merely report the rectoanal inhibitory reflex as being either present or absent. This study aimed to assess the parameters of the rectoanal inhibitory reflex in incontinent and constipated patients and healthy control subjects, in an attempt to analyze differences in internal anal sphincter function in these groups. We analyzed each response of the internal anal sphincter to rectal distention with progressively increasing volumes of air at a single site (proximal anal canal). METHODS: Fifty-five constipated and 99 incontinent patients and healthy control subjects underwent manometry. Various parameters of the rectoanal inhibitory reflex were analyzed, and percentage sphincter relaxation was calculated at each volume at which rectoanal inhibitory reflex occurred. RESULTS: There was no difference in the volume of rectal distention required to elicit sensation (P = 0.626) or the rectoanal inhibitory reflex (P = 0.371) in the three groups. There was a significant correlation between the volume required to elicit the rectoanal inhibitory reflex and that at which sensation was first felt only in the incontinent (P = 0.0001) group. Significantly greater sphincter relaxation was seen at each volume (P = 0.001) in the incontinent as compared with the constipated patients. With progressive rectoanal inhibitory reflex, consistently progressive increases in internal anal sphincter relaxation were found only in the incontinent group. This consistent relationship was not seen in the constipated patients or in healthy control subjects. CONCLUSIONS: Assessment of various parameters of the rectoanal inhibitory reflex yielded important information regarding the continence mechanism. Altered responses of the internal anal sphincter in anorectal disorders plays a role in the associated physiologic impairment. This may have significant clinical implications with regard to sphincter-saving resections.

Vaginography—Investigation of choice for clinically suspected vaginal fistulas

PURPOSE: Vaginal fistulas are rare but can cause extremely distressing symptoms for patients and prove difficult to define anatomically. Barium studies have been reported as having a maximum sensitivity of only 34 percent for detection of vaginal fistulas. Vaginography is an alternative method for diagnosis and evaluation of suspected vaginal fistulas, which has been reported to have a sensitivity of 100 percent. We reviewed our total experience of vaginography to fully assess its capabilities. METHODS: Twenty-seven patients with clinically suspected vaginal fistulas were investigated with vaginography during a six-year period. Results of vaginograms were compared with final operative or clinical diagnosis and with results of other radiologic investigations. RESULTS: Vaginography successfully identified 19 of 24 fistulas, giving a sensitivity of 79 percent. In our series, barium enema was only able to identify 9 percent of fistulas arising from the colon. CONCLUSIONS: In this, the largest series of vaginograms, apparent reduction in sensitivity from the 100 percent quoted in earlier series to 79 percent probably represents a more accurate assessment of vaginography as a diagnostic investigation. Even allowing for this reduction, vaginography is still the most sensitive, economic, and informative investigation for identification and delineation of vaginal fistulas. We recommend that vaginography be the initial investigation of choice in patients with clinically suspected vaginal fistulas.

Randomized clinical trial of Entonox®versus midazolam–fentanyl sedation for colonoscopy†

Intravenous sedation for colonoscopy is associated with cardiorespiratory complications and delayed recovery. The aim of this randomized clinical trial was to compare the efficacy of Entonox® (50 per cent nitrous oxide and 50 per cent oxygen) and intravenous sedation using midazolam–fentanyl for colonoscopy.

Do routinely measured delivery variables predict anal sphincter outcome

PURPOSE: Trauma to the anal sphincter is a recognized complication of primiparous childbirth. This damage may be compounded during subsequent deliveries leading to symptoms. Earlier work is inconclusive as to which delivery variables are associated with such damage and may prove useful in predicting its occurrence, thereby allowing the potential for intervention in these later deliveries to protect the traumatized anal sphincter. The purpose of the present study was to determine whether routinely recorded obstetric variables can be correlated to anal sphincter damage in a consecutive series of females. METHODS: A prospective study was undertaken in a single maternity unit. Patients delivering were assessed before discharge using a symptom questionnaire and endoanal ultrasound. Delivery data were collected prospectively and analyzed statistically to see if a significant difference existed in the presence of an anal sphincter defect. RESULTS: A total of 159 patients were assessed. Endosonography revealed sphincter injuries in 8.7 percent of the normal vaginal delivery group and 83 percent of the forceps delivery group. No correlation was found between head circumference, baby weight, maternal body mass index, epidurals, episiotomy, length of each stage of labor, and duration of active pushing. Forceps delivery was the only factor to be significantly associated with sphincter trauma. CONCLUSION: Besides forceps delivery, commonly measured delivery variables are not useful predictors of anal sphincter trauma. Normal vaginal deliveries do not warrant routine postnatal anorectal assessment. but this should be routine for all instrument deliveries.