Unpublished Scientific Article
Cindy M Mosbrucker MD
David B Redwine MD
2190 NE Professional Court
Bend, Oregon 97701
The age-stratified laterality of distribution of endometriosis refutes the peritoneal circulation modification of the theory of reflux menstruation as the origin of endometriosis.
Objective: To tabulate the frequencies of involvement of left and right sided uterosacral and ovarian endometriosis in age-group intervals.
Design: Retrospective review of contemporaneously-tabulated data.
Setting: Tertiary international referral practice specializing in the surgical treatment of endometriosis.
Patients: All patients with no previous surgical treatment of endometriosis.
Interventions: Pelvic mapping of anatomic sites involved by endometriosis.
Outcome measures: Tabulation of unilateral uterosacral and ovarian disease.
Results: There was a small and mostly statistically insignificant predominance of left sided lesions that did not diverge with increasing age.
Conclusions: The incidence of left-sided disease did not increase with advancing age as would be predicted by the peritoneal circulation modification of Sampson’s theory of origin of endometriosis.
The origin of endometriosis has been the subject of intense speculation for decades. Sampson’s theory of reflux menstruation is favored by many researchers and clinicians. This theory asserts that during menstruation, viable endometrial cells or small endometrial tissue fragments pass from the uterine cavity, through the fallopian tubes, and exit the fimbriated ends of the tubes into the peritoneal cavity. Once these viable cells enter the peritoneal cavity, three sequential steps occur: 1. Viable cells or tissue fragments must attach to intraperitoneal surfaces; 2. These cells or tissue fragments must then proliferate and invade the surfaces to which they have attached; 3. The autotransplant disease called endometriosis becomes established. There is universal agreement that the only acceptable scientific proof of this theory would be robust photomicrographic evidence of the first two steps occurring in vivo. Since such required scientific proof of Sampson’s theory is absent, support for this theory comes from circumstantial evidence.
A recent argument offered as circumstantial evidence of the theory of reflux menstruation is the peritoneal circulation modification of Sampson’s theory. This modification is based on mapping of the frequency of involvement of anatomic sites by endometriosis, such as the uterosacral ligaments and ovaries (1,2,3,4), ureter (5), intestines (6), and sciatic nerve (7). In this type of study, it has been observed among women with surgically diagnosed endometriosis that when only one uterosacral ligament or one ovary is involved, the left side is more commonly involved than the right side, whereas the right sciatic nerve is involved more commonly than the left. It is argued that these differences in laterality of involvement of pelvic structures are due to the alteration of a clockwise flow (as an observer faces a patient) of peritoneal fluid by the sigmoid colon. In this model, a putative peritoneal circulation of fluid descending along the left abdominal wall is directed medially by the presence of the sigmoid colon. Endometrial cells or tissue fragments entering the pelvis via reflux menstruation from the left fallopian tube are thus alleged to remain in a protected microenvironment, trapped in an eddy lateral to the sigmoid colon. This allows these cells to remain in contact with the left side of the pelvis for a longer length of time since they not swept away by the current, resulting in an increased likelihood of attachment, proliferation and invasion and subsequent development of endometriosis on the left side of the pelvis. Each month, when reflux menstruation takes place, the left side of the pelvis is preferentially seeded by the refluxed cells or tissue fragments which are allowed to remain in contact with pelvic surfaces for a longer length of time. The end result of this model is that the left uterosacral ligament and left ovary are more commonly involved than their contralateral twin structures.
It is further argued that the sigmoid colon simultaneously protects the area of the roots of the sciatic nerve on the left side, thus protecting them from reflux menstruation. This concept of an organ simultaneously promoting and inhibiting the occurrence of a disease on the same side of the body is unique in medicine. Since the area of the right sciatic nerve roots is not protected by the sigmoid colon, endometrial cells or tissue fragments refluxed from the right tube can attach to the area of these nerve roots, thus explaining the greater frequency of right-sided sciatic nerve endometriosis according to this theory.
The pelvic mapping studies offered in support of the peritoneal circulation modification of Sampson’s theory have looked at the cumulative frequency of involvement of pelvic sites of interest in populations of women with endometriosis. Most studies have concentrated on the unilateral involvement of these sites, especially the uterosacral ligaments and ovaries.
There is more information to be obtained from pelvic mapping than such simple overall observations of the distribution of unilateral disease in a large group of women.
Age-stratified pelvic mapping would allow new insight into the validity of the pelvic circulation/sigmoid obstruction modification of Sampson’s theory. If with every menstrual flow new endometrial cells were deposited and preferentially allowed to attach and propagate into lesions of endometriosis, then as women get older the left side of the pelvis should be progressively more commonly diseased than the right side of the pelvis. As a result, as older age groups of women with endometriosis are examined, the difference between left and right-sided pelvic involvement should progressively increase. This paper examines the frequency of unilateral endometriosis involving the uterosacral ligaments and ovaries by five-year age intervals to determine if a significant difference of laterality of involvement exists and to determine if this difference progressively increases with advancing age.
Materials and Methods
The computerized database of the Endometriosis Institute of Oregon contains detailed, prospectively-obtained information on over 2,800 women with surgically-diagnosed endometriosis operated on by the senior author (DBR). This is the largest such database in the world, and the details of its management have been described in previous publications (8,9,10).
In addition to administrative information on each patient such as name, birthdate, address, etc, an important component of each patient record is mapping of sites of involvement by biopsy-confirmed endometriosis. Excision of endometriosis has been performed exclusively, with all reported visual manifestations of endometriosis subject to aggressive excision. The pelvic and intestinal sites of involvement by biopsy-proved endometriosis were then entered into fields in the database according to individual anatomic sites of involvement as previously described (11). Institutional review board approval for this study was not sought since pelvic mapping did not affect patient treatment, did not violate patient confidentiality, and is an extension of the revised American Fertility Society classification system of endometriosis, which is a recommended routine part of patient record-keeping.
The database was queried for patients with no previous surgical therapy who had been found to have histologically documented endometriosis of the uterosacral ligaments and ovaries. Patients were stratified into five-year age intervals.
Statistical analysis was performed using McNemar’s chi-square analysis to accept or exclude the null hypothesis that there is no significant difference in laterality of disease for each age interval. A significance level of p<0.05 was chosen in order to match most previous publications on this topic. A significance level of p>0.05 would therefore indicate that the difference in involvement of a left sided pelvic structure versus its twin on the right side was not statistically significant. Also, for each age interval, the amount (positive or negative) by which the frequency of left sided disease exceeded that of the right side was observed to see if it remained positive across all age groups and increased progressively across advancing age groups, which would be an inevitable result of the peritoneal circulation modification of Sampson’s theory.
A total of 2841 patients were in the database, 1471 of whom had no previous surgical therapy for endometriosis or pelvic pain. Of these, 845 patients had biopsy proven endometriosis involving either one or both uterosacral ligaments. Overall, 224 had only left sided disease, and 165 had only right sided disease, with 456 having bilateral disease of the uterosacral ligaments.
When patients with unilateral disease of the uterosacral ligaments were stratified into eight five-year age intervals (Table 1), five of the eight age intervals had disease more commonly found on the left side, although this finding was statistically significant only for the 25 – 29 year age interval. When the chi square analysis was performed for the entire group of patients with uterosacral ligament disease, the chi square level was 8.95, p <0.005. However, this significance for the entire population was due entirely to the effect of the findings in the 25 – 29 year age interval. The difference between the frequencies of left and right sided disease of the uterosacral ligaments did not increase sequentially with, instead varying positively or negatively among the age groups.
Four hundred and three patients had biopsy proven endometriosis of the ovaries, with 130 involving the left ovary only, 122 the right ovary only, and 151 with bilateral ovarian endometriomas, a non-significant difference in laterality. Four of the eight age group intervals had ovarian disease more commonly found on the left side than on the right. None of these differences was statistically significant, and an increasing frequency of involvement of the left ovary vs the right ovary was not seen with advancing age. As with uterosacral ligament involvement, the differences between the frequencies of left and right-sided ovarian endometriosis varied positively and negatively across the age intervals and did not show a progressive increase with age as the peritoneal circulation model predicts.
The peritoneal circulation modification of Sampson’s theory of origin of endometriosis is refuted by these findings. The laterality of involvement of pelvic structures cannot be used as circumstantial evidence for Sampson’s theory.
Many authors have described trends related to laterality of endometriosis lesions involving various organs and structures. The slight left-sided predominance has been championed as strong circumstantial evidence in favor of Sampson’s theory. These authors commonly cite studies which they believe establish the existence of a peritoneal circulation that supposedly runs clockwise as an observer looks at a patient. If this notion were true, then each month with the menstrual flow, endometrial cells or tissue fragments should preferentially accumulate in areas protected from this circulation, such as the left side of the pelvis which allegedly is protected by the sigmoid colon. Since this process is argued to begin with menarche, progressively older age groups of women should have progressively more disease in those protected areas. This is because each month refluxed cells are preferentially allowed to remain in longer contact with pelvic surfaces in such a protected area, allowing the cells an increased opportunity to attach and then to proliferate and invade. With each passing menstrual flow, as more disease accumulates in such a susceptible protected location, the difference in frequency of involvement of the left vs right side of the pelvis should increase as older age groups are examined. Our results indicate clearly that this does not occur.
Our age-stratified results forcefully contradict the notion that the frequency of involvement of the left side of the pelvis is statistically different from the right side, since such statistical significance was found in only one age group of patients with unilateral uterosacral ligament disease and in none of the age groups with ovarian disease. Additionally, our results indicate that an increasing frequency of involvement of the left side of the pelvis does not occur with advancing age, a direct refutation of one of the inevitable predictions of the simplistic peritoneal circulation modification of Sampson’s theory. These results were revealed by age-stratified analysis of patients with endometriosis, illustrating the imprecision and misleading results which can result from collapsing all data into one large data set as has been done with most previous papers.
One other paper presented results with findings similar to ours. Bazi et al (3) found that patients less than 35 years of age had a statistically significant predilection of endometriomas for the left ovary than patients 35 and older, although the numerical basis for their calculation was not presented and patients less than 26 years of age were not included in the analysis.
A two-part question arises in considering the alleged peritoneal circulation: What evidence has been published which supports the existence of this peritoneal circulation in the first place, and does this evidence support the contention of champions of the peritoneal circulation modification of Sampson’s theory? The literature was searched for papers which might shed light on the existence of the type of peritoneal circulation alleged to exist by supporters of the peritoneal circulation modification of Sampson’s theory.
Mitchell (12) studied the patterns of spread of intraperitoneal fluid in a rather unphysiologic way. Working on stillborn infants, artificial perforations were created in the intestinal tract and barium was injected in to the intestines over three hours, during which x-rays were taken. Depending on where the perforation was created, barium could flow literally in any direction. In one female stillborn, he injected barium into the uterus at 100 mm Hg pressure and noted that the barium flowed from the right tube up the right colonic gutter toward the right subphrenic space. He mentioned this single instance of intrauterine injection in his paper, but did not include this in his discussion, since he considered this single observation to be unphysiologic due to the high pressure of injection. The methodology in this paper is unphysiologic and cannot be accepted as what happens in living adult females. However, if the evidence observed during the single case of intrauterine injection were to be accepted as valid, it would predict that involvement of the right colonic gutter and right upper quadrant would be more frequent than involvement of the right hemipelvis, and this does not occur.
Drye (13) found that in the standing position, the hydrostatic pressure in the bottom of the human pelvis was higher than the hydrostatic pressure in the epigastrium as measured by balloon catheters. When subjects were upright, this pressure difference remained constant, other than slight respiratory variations. This pressure difference went away when the subjects were supine. These results are unsurprising and straightforward manifestations of basic physics. The increased pressure in the pelvic region in the standing position is not evidence of a peritoneal circulation, nor would it force fluid out of the pelvic region into the upper abdomen any more than fluid in the bottom of a glass would be forced to the surface by the pressure difference alone. The unvarying difference of pressure between the upper abdomen and lower abdomen is proof that a circulation does not exist in the upright position.
Meyers in 1970 (14) studied the distribution of radioopaque dye injected into the peritoneal cavity in 20 patients with intraperitoneal effusions and abscesses. Dye was injected into the abdomen through the left upper quadrant. This was done in various positions on a tilt table but not in the upright position. Fluoroscopy revealed the dye to initially sink into the pelvis, then rise up both colonic gutters as the table was tilted, with more pronounced upward flow on the right than the left. The fluid ascending the right gutter pooled preferentially in Morrison’s pouch beneath the liver, from where it could then drift into the sub-phrenic space on the right. This study by Meyers was rather unphysiologic since patients had intraperitoneal pathological conditions, and the patient’s position was changed at will on a tilt table. Also, the observations during the short duration of a radiological exam would not necessarily be an accurate indication of what happens over several years in vivo. For these reasons, this study cannot be accepted as evidence of a natural peritoneal circulation as envisioned by supporters of the peritoneal circulation modification of Sampson’s theory. However, for the sake of argument, if this study by Meyers were accepted as proving that a natural peritoneal circulation exists, then it could not be characterized as clockwise since fluid flowed up the left colonic gutter also. The fact that dye ascended from the left side of the pelvis up the left paracolic gutter means that there is not a “protected microenvironment” on the left side of the pelvis due to the presence of the sigmoid colon. Therefore, the inference that the sigmoid colon offers a protected environment on the left side of the pelvis is false, based on Meyers’ findings. If the peritoneal circulation modification of Sampson’s theory were true, there should be some substantial involvement of the left colonic gutter in the lower abdomen due to the flow upward along the left pelvic and abdominal sidewall, but disease in that area is rare. Also, involvement of Morrison’s pouch by endometriosis would be more common than diaphragmatic disease since particles of endometrium should be carried there first, yet involvement of Morrison’s pouch by endometriosis is extremely rare, if in fact it has ever been reported. Therefore, sites of pelvic and abdominal involvement by endometriosis are not accurately predicted by the patterns of artificially induced peritoneal flow described in this study.
A subsequent study by Meyers (15) correlated sites of metastatic cancer with the flow of intraperitoneal dye. Again, dye was injected into the abdomen and x-rays were taken in multiple positions, including Trendelenberg and oblique views. Up to 200 mL of dye was injected, and the hypertonicity of the dye induced mild ascites, which may have altered the results. The dye collected in the cul-de-sac, the termination of the small bowel mesentery (terminal ileum), the superior aspect of the sigmoid mesocolon, and the right paracolic gutter. These findings were correlated with metastatic implants from ovarian, pancreatic, gastric, and colon cancer. Not surprisingly, among 15 patients with ovarian cancer (the closest approximation to what is alleged to result from reflux menstruation) nine had seeding of the cul-de-sac, 7 had seeding of the small bowel mesentery near the ileocecal junction, 1 had seeding of the medial aspect of the sigmoid mesentery, and 4 had seeding of the right colonic gutter. Once again, the nature of the investigation was unphysiologic due to the tilting of the table during the short duration of the exam, and no clockwise circulation of fluid was expressly described, so this paper cannot be accepted as evidence of a clockwise circulation. However, if the results of this study were to be accepted as predicting where endometriosis would likely occur by virtue of Sampson’s theory, endometriosis should occur preferentially in the cul-de-sac, small bowel mesentery, medial sigmoid colon, and right colonic gutter, following the patterns seen with ovarian cancer. While endometriosis is most commonly found in the cul-de-sac, it rarely involves the small bowel mesentery near the ileocecal junction, and involvement of the medial sigmoid colon and right colonic gutter are uncommon. Once again, sites of pelvic and abdominal involvement by endometriosis are not accurately predicted by the patterns of peritoneal flow described in this study.
Rosenshein et al (16) conducted a study in anesthetized rhesus monkeys where two catheters were inserted into the abdominal cavity. One catheter was inserted in the right upper quadrant and advanced down into the pelvis. The second catheter was inserted in the left upper quadrant and not advanced. Radioactive albumen was injected through the left upper quadrant catheter in a volume of 250 mL in three animals and 10 mL in two animals. Over the course of almost three hours of observation, various maneuvers were applied to the monkeys, including abdominal massage for five minutes, and various position changes including 30 degree Trendelenburg and reverse Trendelenburg positions. The fluid was noted to settle in the cul-de-sac and eventually track up the course of the right-sided catheter when 250 mL was injected, with less pronounced distribution of the smaller volume of fluid. This finding is unsurprising since it demonstrates that fluid will take the path of least resistance, unphysiologic since menstruating women don’t usually have such catheters in place and probably don’t have 250 mL of refluxed menses in their peritoneal cavity, and therefore unhelpful as evidence supporting either a clockwise peritoneal circulation or Sampson’s theory. The authors of this study said nothing about a clockwise circulation existing in the peritoneal cavity of rhesus monkeys.
Rosenshein et al, (17) in a discussion of the diffusion of radiocolloids in the peritoneal cavity describes the actions of the diaphragm as a ‘pump, draining the peritoneal fluid upwards with each inspiration.’ There is nothing in the context of the rest of this paper which would lead anyone to conclude that there is a clockwise or counterclockwise current in the abdominal cavity.
Foster et al (18) published a paper commenting on the right-sided predilection of thoracic endometriosis. Citing Rosenshein et al (17), they presented a schematic diagram of a clockwise circulation of peritoneal fluid, although they presented no evidence of their own to suggest that this circulation existed. This schematic diagram, a crude and incomplete explanation of Rosenshein’s findings, cannot be accepted as scientific evidence of a clockwise peritoneal circulation.
McQueen et al (19) placed 1.5 mL of radionuclide in the vaginas of 23 women without endometriosis and found that after one hour, the material had collected in the cervix, uterus, and around the adnexae bilaterally. A peritoneal circulation was not being sought and such a circulation was not mentioned in this paper.
From this review of the literature it is clear that there is no scientific evidence of the existence of a clockwise peritoneal circulation which is alleged to exist by proponents of the peritoneal circulation modification of Sampson’s theory. Supporters of this theory do not cite any reference with scientific evidence of this circulation. Any arguments which are based on the alleged existence of such a circulation are therefore without foundation.
The peritoneal circulation modification of Sampson’s theory is overly simplistic and does not address fundamental questions which arise from it. For example, is the alleged peritoneal circulation a stronger force on the tissue fragments than that exerted by surface tension of moist surfaces? Are the effects of gravity stronger than the alleged circulation?
The statistics which have been used to support the peritoneal circulation modification of Sampson’s theory are inconsistent with our findings and occasionally internally contradictory. Most of the published papers on the laterality of endometriosis have focused on the cumulative frequencies of involvement of the uterosacral ligaments and ovaries, since these are distinctly lateral structures about which there would be no confusion over assignment of laterality. Most of the papers on the subject have examined the relative left vs right laterality of involvement of paired pelvic structures by McNemar’s chi square test (although this was not specifically mentioned in the Methods sections of these papers.). The schematic formula for this statistic is (L – R)2/T, where L is the number of patients with left-sided unilateral disease and R is the number of patients with right-sided unilateral disease and T is the sum of L + R.
One paper published in support of the peritoneal circulation modification of Sampson’s theory actually contains statistical evidence refuting that theory. Parazzini, et al, (20) wrote a paper comparing the incidence of endometriosis in the left vs. right hemipelvis. Rather than using McNemars’s chi square test (which was used in all other papers published previously on this subject by members of this writing group and numerous other authors) to analyze their data, they instead used the binomial distribution and Poisson’s approximation to measure the significance of their findings that the left side was more commonly involved than the right side. However, when the traditionally-used McNemar’s chi-square test is applied to their data, the chi-square value is 2.58 (p>0.1) which is not significant. Thus, this paper actually refutes its authors’ claims when the correct statistical test is run. In order for a theory to be valid, it should be consistently supported by the same statistical test used to support it in other publications rather than searching for a different statistical test which might give support.
Papers championing the peritoneal circulation modification of Sampson’s theory have avoided the question of the large percentage of patients with bilateral disease. For a theory to be valid, it must include an explanation for all observations, not just a selected portion such as those with only unilateral disease.
Reflux menstruation itself as a theory of origin of endometriosis remains unproven, supported only by many lines of circumstantial evidence. The peritoneal circulation modification of this theory has introduced a new, strained attempt to support this theory by circumstantial evidence. If such a peritoneal circulation were to exist, our findings would suggest that this circulation reverses direction at various times in a woman’s lifespan, then eventually slows. Such variation of an unproven circulation seems unlikely and has not been proposed by supporters of this concept.
The peritoneal circulation modification of Sampson’s theory of reflux menstruation as the origin of endometriosis has resulted in an unusual situation in scientific publication. A long-standing but scientifically unproven theory of origin (Sampson’s theory) of an important disease seeks support from inconsistent and contradictory statistical evidence of laterality of disease which infers the physical results of a peritoneal circulation which has not been scientifically proven to exist. Since the peritoneal circulation modification of Sampson’s theory is based on a scientifically unproven fiction, it should be abandoned.
There is a slight left sided predominance of endometriosis involving the uterosacral ligaments and ovaries. The difference between the frequency of left-sided involvement and right-sided involvement does not increase with advancing age. The frequency of involvement of the left side of the pelvis does not increase linearly with advancing age. A clockwise circulation of peritoneal fluid has not been scientifically proven to exist. Supporters of the peritoneal circulation modification of Sampson’s theory have selected portions of studies which support their contentions and have ignored those portions that do not. The laterality of distribution of endometriosis does not support Sampson’s theory via the peritoneal circulation model. Future papers on the theory of reflux menstruation as the origin of endometriosis should focus on the most easily obtainable scientific evidence of all: robust photomicrographic evidence of the in vivo occurrence of the two missing steps of this theory, adhesion and implantation of the endometrial cells at their final resting sites within the pelvis.
- Vercellini P, Busacca M, Aimi G, Bianchi S, Frontino G, Crosignani PG. Lateral distriburion of recurrent ovarian endometriotic cysts. Fertil Steril 2002;77:848-9.
- Vercellini P, Aimi G, De Giorgi O, Maddalena S, Carinelli S, Crosignani PG. Is cystic ovarian endometriosis an asymmetric disease? Br J Obstet Gynaecol 1998;105:1018-21.
- Bazi T, Nader KA, Seoud MA, Charafeddine M, Rechdan JB, Zreik TG. Lateral distribution of endometriomas as a function of age. Fertil Steril 2007;87:419-21.
- Al-Fozan H, Tulandi T. Left lateral predisposition of endometriosis and endometrioma. Obstet Gynecol 2003;101:164 – 6.
- Vercellini P, Pisacreta A, Pesole A, Vicentini S, Stellato G, Crosignani PG. Is ureteral endometriosis an asymmetric disease? Br J Obstet Gynaecol 2000;107:559 – 61.
- Vercellini P, Chapron C, Fedele L, Gattei U, Daguati R, Crosignani PG. Evidence for asymmetric distribution of lower intestinal tract disease. Br J Obstet Gynaecol 2004;111:1213-7.
- Vercellini P, Chapron C, Fedele L, Frontino G, Zaina B, Crosignani PG. Evidence for asymmetric distribution of sciatic nerve endometriosis. Obstet Gynecol 2003 102:383-7.
- Redwine DB. Database management for personal computers, I. J Am Assoc Gynecol Laparosc 1996; 3: 469-74.
- Redwine DB. Database management for personal computers, II. J Am Assoc Gynecol Laparosc 1996; 3: 643-50.
- Redwine DB. Database management for personal computers, III. J AmAssocGynecol Laparosc 1996; 4: 95 – 101.
- Redwine, DB. The distribution of endometriosis in the pelvis by age groups and fertility. Fertil Steril 1987; 47:173-175
- Mitchell GAG. The spread of acute intraperitoneal effusions. Br J Surg 1941;28:291 – 313.
- Drye JC. Intraperitoneal pressure in the human. Surg Gynecol Obstet 1948;87:472-5.
- Meyers MA. The spread and localization of acute intraperitoneal effusions. Radiology 1970;95:547 – 554.
- Meyers MA. Distribution of intra-abdominal malignant seeding: Dependency on dynamics of flow of ascitic fluid. Am J Roentgenol Radium Ther Nucl Med 1971;119:198 – 206.
- Rosenshein N, Blake D, McIntyre PA, Parmley T, Natarajan TK, Dvomicky J, et al. The effect of volume on the distribution of substances instilled into the peritoneal cavity. Gynecol Oncol 1978;1:106 – 10.
- Rosenshein NB, Leichener PK, Vogelsang G. Radiocolloids in the treatment of ovarian cancer. Obstet Gynecol Surv 1979;34:708 – 20.
- Foster DC, Stern JL, Bescema J, Rock JA, Woodruff JD. Pleural and parenchymal pulmonary endometriosis. Obstet Gynecol 1981; 58:552 – 6.
- McQueen D, McKillop JH, Gray HW, Callaghan M, Monaghan C, Bessent RG. Radionuclide migration through the genital tract in infertile women with endometriosis. Hum Reprod 1993;8:1910 – 4.
- Parazzini F, et al. Left:right side ratio of endometriotic implants in the pelvis. Eur J Obstet Gynecol 111(2003) 65-67.
Endometriosis, theories of origin, laterality, pelvic mapping