REVIEW: Evidence for asymmetric distribution of lower intestinal tract endometriosis
The aetiology of endometriosis is controversial. Investigating the anatomical distribution of endometriotic lesions may provide insights into the pathogenesis of the disease.1 If ectopic endometrium is due to retrograde menstruation, the pattern of lesions should be determined mainly by anatomical and physiological variables, whereas if coelomic metaplasia is the cause of endometriosis, lesions should not be distributed in relation to factors influencing the spreading and implantation of endometrial cells.1,2
The pattern of involvement of bilateral and symmetric pelvic organs has been studied specifically to verify whether endometriotic lesions are equally distributed on the two sides. In this regard, it has been demonstrated that the left ovary, ureter, and uterosacral ligament, as well as the right inguinal canal structures and sciatic nerve are affected significantly more frequently than the contralateral corresponding organ.2–12 This preference for the left side constitutes indirect evidence against the coelomic metaplasia theory, which is more likely to be associated with equal distribution, and the laterality has been attributed to the presence of the sigmoid colon on the left hemipelvis.2–12
The lower intestinal tract is the most common site of extragenital endometriosis. This is not unexpected, given its proximity to the fallopian tubes.13 However, if the menstrual reflux theory is true, the left terminal colon should be affected more frequently than the right one. In fact, the sigmoid is in close contiguity to the ipsilateral tube and creates a sort of shelter which should facilitate implantation of regurgitated endometrial cells, whereas the caecum is more cranial and hence less prone to be involved.
It has been observed that endometriotic lesions of the descending colon are more common than those of the ascending colon,14 but the issue has never been approached formally. Moreover, the evidence in this regard is confusing, because endometriosis of the rectal tract is often considered among lesions of the left side. This may not be true, as the relatively frequent involvement of the rectum could originate from implants in the pouch of Douglas and be considered a primarily midline lesion.15 Finally, deep, infiltrating and obstructive lesions were not always distinguished from superficial, non-obstructing implants.
To verify the hypothesis of an asymmetrical distribution of endometriotic lesions of the ileocaecum and colon, we reviewed all data published on the topic in the English-language scientific literature in the last two decades. A detailed description of the patients' symptoms and signs and of the diagnostic and treatment alternatives was not among the aims of the study.
Several different strategies were adopted to identify all English-language medical papers on intestinal endometriosis published from 1980 to 2003. We conducted a MEDLINE, EMBASE and PUBMED search using combinations of medical subject heading terms: endometriosis, ileum, colon, caecum, appendix and sigmoid. All pertinent articles were retrieved and additional reports were then identified by systematically reviewing all references. In addition, books and monographs on endometriosis published in the last 15 years were consulted. Proceedings of scientific meetings were not included.
We considered only articles in which the presence of a deep infiltrating endometriotic lesion of the terminal ileum and colon required intestinal resection. Two authors (U. G. and R. D.) abstracted data in an unblinded fashion on standardised forms. An initial screening of the title and abstract of all articles was performed to exclude citations deemed irrelevant by both observers (i.e. when only superficial implants were treated or when lesions were found on the proximal ileum or transverse colon). Endometriosis of the terminal ileum (within 15 cm of the ileocaecal valve), caecum and ascending colon was considered a right-side lesion, that of the descending and sigmoid colon a left-side lesion. Lesions of the transverse colon and of the rectum, considered as midline disease, were specifically excluded.
Appendiceal lesions were included when associated with deep endometriosis at other intestinal sites. Isolated endometriosis of the appendix was excluded because the vast majority of cases were retrieved from large appendectomy series in which histological demonstration of the disease was incidental or sought retrospectively. In this type of study, the surgical indication was unclear and the exploration of the entire colon not reported or not reliable. Moreover, in the above studies as well as in most of the published case reports, it is not possible to distinguish between superficial and deep infiltrating appendiceal lesions. Finally, appendectomy may not be considered as a true bowel resection.
The year of publication, clinical characteristics of subjects, results of pre-operative diagnostic investigations and surgical details were recorded independently. The number of women with endometriosis of the ileocaecum and colon and the side of the lesion were obtained from individual studies.
The combined frequency of left- and right-sided lower intestinal tract endometriosis in published reports was analysed with the χ2 test to compare observed and expected events. The confidence interval (CI) of the proportion of endometriosis of the left colon was computed using the normal approximation.
A total of 68 studies were initially identified by computerised database searches as potentially relevant10,14,16–81 and a further eight were identified by hand-searching and by checking bibliographies.82–89 Eight of the reports were excluded for various reasons: four because the exact sites and/or number and/or type (superficial vs deep infiltrating) of bowel endometriotic lesions were not specified32,43,52,65; three because endometriotic lesions at different sites not affecting the terminal ileum and colon were described10,45,71; and one because diagnostic findings were reported without details of subsequent surgery and histologic examination.79 In 22 articles,14,18,20–22,25,30,31,34,46,50,53,56,62–64,70,81,82,86,87,89 773 women affected by various lesions were described, but only the 156 with colic resection were included in the analysis, excluding the subjects who underwent surgery for rectal (n= 139), isolated appendiceal (n= 47) and superficial only (n= 205) lesions, those with non-endometriotic intestinal pathology (n= 197) and those with endometriosis not involving the bowel (n= 26). Three other patients were excluded for various reasons (intestinal site not specified, 1; lesion of the transverse colon, 1; surgery refused, 1). One case of bilateral bowel resection for endometriosis was described in eight articles.25,27,31,35,38,46,64,70
Sixty-eight reports published in the English literature were finally included. A total of 337 subjects were selected, 333 with histopathological demonstration of intestinal endometriosis, and 4 in whom surgical exploration and intestinal resection were performed, but no pathological diagnosis was reported formally.
The median number of cases observed was 5 (range, 1–77), and the median age of the subjects 33 years (range, 19–78). Pelvic endometriotic lesions, in addition to those involving the bowel, were observed in 111/337 (33%) cases. Endometriosis involved the terminal ileum in 66 cases, appendix in 13, caecum in 34, ascending colon in 2, descending colon in 4 and sigmoid in 252 (proximal, 63; distal, 138; unspecified, 51). The sum (n= 371) exceeds the total because 26 women had more than one intestinal lesion.
Endometriosis of the resected colon was on the right side in 84 patients, on the left in 245 and bilateral in 8. Considering only the patients with unilateral colonic endometriosis, the observed proportion of left-side lesions (245/329, 74%; 95% CI 69–79%) was significantly different from the expected proportion of 50% (χ21, 78.8, P= 0.0001). Of the four subjects who underwent resection for deep endometriosis of the ileocaecum and colon without reported histopathological confirmation of intestinal involvement, three had left-sided lesions and one bilateral lesion. Exclusion from the analysis of these four women did not modify the overall results.
To our knowledge, no systematic review of lower intestinal tract endometriosis has yet been published, and for this reason we analysed all available data on the topic identified by means of a thorough review of the English literature of over two decades. Various article search strategies were adopted. Data were abstracted from standardised forms compiled by two independent investigators, who admittedly were not blinded. Rejected studies and the reasons for their exclusion are described.
Our results confirm that endometriosis of the lower intestinal tract, like that of the ovary, uterosacral ligament and ureter, develops significantly more frequently on the left than on the right side. Interestingly, the proportion of left-sided gonadal, uterosacral ligament and ureteral lesions is remarkably similar (63%, 65% and 64%), whereas that of left bowel lesions is higher (74%), confirming previous observations.13,14
Unexpectedly, pelvic endometriotic implants, in addition to bowel lesions, were described in only one-third of the subjects. However, this finding must be considered with caution because abdominal surgeons may be less concerned than gynaecologists in describing genital endometriosis in detail.
In the present review, the number of patients with bilateral intestinal lesions is small (8/337, 2.4%). Admittedly, bilateral bowel involvement could have been under-estimated as in some studies it was not possible to determine whether lesions were uni- or bilateral, because they were simply listed without reference to possible associations in the same subject. However, the potential erroneous computation of a bilateral lesion as both a right- and a left-sided lesion should have reduced the difference in proportion in favour of the left side.
The coexistence of a left ovarian endometrioma is not consistent or not always verifiable in the cases reported in the literature. However, endometriosis of the descending colon and sigmoid may not necessarily be secondary to endometriotic cysts, but more generally to ectopic implantation of endometrial cells along the left paracolic gutter and lateral intestinal wall. In fact, the left adnexal region is covered by the sigmoid colon, as this portion of the large bowel leans on the left tube and ovary and is very often fixed to the pelvic brim by filmy adhesions. A ‘protected’ microenvironment is established around the left adnexa and, as a consequence, adhesion and implantation on the lateral aspect of the sigmoid of endometrial cells regurgitated through the left tube may be facilitated. The caecum is more cranial and probably less likely to be the site of implantation of endometrial cells regurgitated through the right tube. The asymmetry of ovarian, uterosacral ligament, ureteral and intestinal forms may be the expression of a common underlying anatomical condition that facilitates adhesion and growth of endometrial cells on the left pelvic side wall.
In our opinion, neither the coelomic metaplasia nor the embryonic cell rest theory explains such a clear-cut difference in frequency distribution of colonic lesions between the two sides. Indeed, our findings are compatible with the menstrual reflux theory and with the anatomical differences between the left and right hemipelvis.
The results of our review demonstrated a lateral asymmetry in the location of lower intestinal tract endometriosis, as almost three quarters of the patients had lesions on the left-hand side. This constitutes further evidence against the coelomic metaplasia theory and supports the notion that endometriosis originates from the endometrium and that implantation of refluxed cells occurs according to anatomical and physiological determinants.1,13,15