1 Wellcome Trust Centre for Human Genetics, University of Oxford, Roosevelt Drive, Oxford OX3 7BN, UK, 2 Department of Human Genetics and Biostatistics, University of Pittsburgh, Pittsburgh, PA, 3 Wisconsin National Primate Research Center and 5 Harlow Primate Laboratory, University of Wisconsin-Madison, Madison, WI, USA and 4 Nuffield Department of Obstetrics and Gynaecology, University of Oxford, Oxford, UK
6 To whom correspondence should be addressed. e-mail: krina.zondervan{at}well.ox.ac.uk
![]() |
Abstract |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Key words: animal model/endometriosis/genetic epidemiology/prevalence/rhesus macaque
![]() |
Introduction |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Endometriosis is regarded as a complex trait, in which multiple gene loci conferring susceptibility interact with each other and environmental factors to produce the phenotype. Studying the disease in women is complicated for many reasons. The need for surgical diagnosis means that patient groups are often highly selected relative to the general population in terms of the environmental, and possibly genetic, background. This necessitates careful selection of control groups so as not to cause spurious environmental and genetic associations, a point that not much attention has been paid to in studies to date (Zondervan et al., 2002). In addition, the influence of environmental exposures is difficult to assess as they should have occurred prior to disease onset, which is likely to have preceded diagnosis by many years (Hadfield et al., 1996
). An animal model such as the rhesus monkey (Macaca mulatta) could be a valuable tool to investigate the genetic epidemiology of endometriosis. Colonies of captive rhesus macaques have smaller gene pools and greater genetic homogeneity, and thus may possess unique or high-frequency risk alleles for endometriosis. Moreover, they live in a controlled and monitored environment, which allows a more accurate retrospective analysis of environmental exposures that may have occurred during the animals life span than is possible in humans (although, as in humans, the age at onset of endometriosis is unknown in the rhesus macaque).
Endometriosis develops spontaneously in the rhesus macaque and the endometrial implants are morphologically identical to its human counterpart. Clinical manifestations include altered behaviour likely related to discomfort, abdominal distension due to a pelvic mass, and occasionally cachexia because of full thickness bowel involvement (MacKenzie and Casey, 1975). The rhesus macaque shares many aspects of its anatomy and physiology with humans, in particular reproductive physiology. Menarche in the rhesus monkey occurs at
3 years of age, the length of the menstrual cycle is
28 days, menstrual bleeding lasts for
4 days and ovarian function diminishes from the age of
25 years onwards. The rhesus macaque is a seasonal breeder in nature (Catchpole and van Wagenen, 1975
), but this seasonality disappears when housed indoors on a constant light/dark and temperature/humidity regimen. The average life span is 2529 years but some animals live to 40 years of age (Colman and Kemnitz, 1998
).
We previously reported on the high prevalence of spontaneous endometriosis in the rhesus macaque colony of the Wisconsin National Primate Research Center (WNPRC) at the University of Wisconsin-Madison (Hadfield et al., 1997). Briefly, necropsy records from 399 female rhesus monkeys that died
4 years of age (the age at which sexual maturity is reached) between 1981 and 1995 were examined for evidence of endometriosis. Eighty-one of the animals (20%) were found to have disease at necropsy. All diseased animals were aged
10 years; prevalence among animals that had reached at least the age of 10 years was 29%. Comparison between 62 cases and 62 controls matched on age and year of death showed that cases were more likely to have been exposed to estradiol (E2) and hysterotomy. Moreover, animals in the case group had a higher sum of kinship coefficients (4.375) than did controls (1.6875), suggesting a higher degree of relatedness. In the present paper, we investigate this familial aggregation further, using pedigree information and incorporating colony information updated since 1995.
![]() |
Methods |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Until 1978, WNPRC and the Harlow Primate Laboratory used one, paper-based, recording system for the animals. Since then, records have been maintained separately. The Harlow Primate Laboratory has continued to use written records in which information on necropsies, surgical and experimental procedures, and familial relationships of animals living in their colony is recorded. From 1981, WNPRC has held computerized records (back-dated to 1978) of the clinical, experimental and environmental histories of all animals for their time of residence in the colony, as well as their familial relationships. Since animals were often moved between WNPRC and the Harlow Primate Laboratory for breeding purposes, the computerized database contains animals that may have lived at either institute. In the remainder of the paper, we will refer to this group of >9000 animals for which data are registered on the database as the colony. After death, all animals undergo necropsy, the records of which are also held on the computerized database. Parentage is known for 90% of animals because of arranged mating and post-birth DNA checks. Inbreeding is avoided by checking the kinship between animals prior to every mating.
Identification of animals with endometriosis in the colony (1978 to September 2001)
Unequivocal information on the presence or absence of endometriosis could only be reliably obtained retrospectively through necropsy records. Surgical records were also searched, but rarely contained information about the presence of endometriosis (see Table I), because the pelvis was not routinely inspected for the condition (although it is fair to assume that severe disease, if present, would have been commented on). Most operations were performed for experimental purposes linked to reproductive medicine/fetal projects.
|
Stage of disease was determined as minimal-mild or moderate-severe according to the rAFS classification system (The American Fertility Society, 1985) using the description contained in the necropsy or surgery records. This staging system has similarly been used in other endometriosis studies of non-human primates (Schenken et al., 1987
; DHooghe et al., 1992
), and its use presented little difficulty because animals tended to have either a few isolated foci or a severely affected pelvis.
Definition of unaffected animals
All affected animals identified in the colony were aged 10 years or older at diagnosis or death. Other female animals in the colony database were only considered unaffected if: (i) they had died at WNPRC or the Harlow Primate Laboratory aged 10 years or older; (ii) they had not had a bilateral ovariectomy >1 year prior to death; and (iii) a full necropsy report was available providing evidence of absence of endometriosis. The criterion of having at least one intact ovary was applied because ectopic endometriotic tissue is not viable in ovariectomized monkeys (DiZerega et al., 1980). Note that lack of endometriosis reported at pelvic surgery did not constitute sufficient evidence for a female to be considered unaffected. Thus, 284 females were diagnosed as unaffected at death between 1978 and September 2001.
Construction of pedigrees (followed up from September 2001 to present)
To investigate the degree of familial clustering of endometriosis, pedigrees were constructed using all 142 cases identified in the whole colony up to September 2001. All founding ancestors of the cases were identified from the colony database, and all female descendants and any inter-related males were included. Since September 2001, only living females which are part of the pedigrees are being followed up to determine disease status at necropsy or surgery. The last update of disease information in the pedigrees, on which the recurrence risk (RR) analyses in the present paper were based, was in June 2002.
Statistical analyses
Differences in age at diagnosis or death between affected and unaffected animals were assessed using a t-test. Prevalence rates of endometriosis in the colony could only be calculated among females in which presence or absence of endometriosis was unequivocal, i.e. those that had died aged 10 years up to September 2001, had not had a bilateral ovariectomy more than 1 year prior to death and for which a full necropsy report existed (n = 414). Significance of the observed rise in prevalence rates with age (in categories) and with calendar year (in categories) was investigated using a
2-test for trend. A multivariate logistic regression model was used to assess the effect of calendar year of diagnosis adjusted for age at diagnosis.
Kinship among all 130 affecteds in the colony diagnosed at necropsy up to September 2001 was compared with that among a random sample of 130 unaffecteds that were frequency-matched to affecteds on age [categories (years): 1015 and 16+] and calendar year of death (categories: up to 1980, 19811985, 19862001). Matching was employed because age and year of death affected both risk of disease and likelihood of having offspring. Using genealogical data from the entire WNPRC colony, the program Kinmean (S.A.Sholl) was used to calculate kinship coefficients for each unique pair within the groups of affected and unaffected animals (Boyce, 1983). For example, a kinship coefficient of 0.25 indicated a full sib or parentchild relationship and 0.125 a half sib or auntniece pair. Differences in the distribution of these values between groups of affected and unaffecteds were compared using the non-parametric MannWhitney U-test. Crude segregation ratios (the proportion of affected daughters among all daughters with known disease status) in the colony were compared between affected and unaffected mothers using a
2-test.
Using pedigree information up to June 2002, RRs (the probability of an affected females relative being affected) were calculated for full sibs and half sibs using the method described by Olson and Cordell (2000). This method calculates RR among sibling pairs adjusting for sibship size when proband status is unknown:
where a = number of affecteds in sibship; wa = weight given to the sibship (in our situation of complete ascertainment of cases wa = a); s = sibship size; and ns(a) = the number of sibships of size s with a affecteds. Recurrence risks for motherdaughter and grandmothergranddaughter pairs were calculated as the proportion of affected (grand)daughters among all (grand)daughters of known disease status given that (grand)mothers were affected. Differences between RRs were assessed using Fishers exact test.
![]() |
Results |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
The prevalence of endometriosis among all females in the colony that died aged 10 years up to September 2001, without bilateral ovariectomy and with a full necropsy, was 130/414 = 31.4% [95% confidence interval (CI) 26.935.9%]. Prevalence of moderate-severe endometriosis only was 98/407 = 24.1% (95% CI 19.928.2%). Prevalence (all stages) increased significantly (Figure 1A) from
10% among those that died aged 1012 years to
40% among those aged 15 years or older at death (
2trend = 23.0; P < 0.001). A similar pattern with age was observed for moderate-severe disease only. Total pedigree prevalence rates also differed significantly with calendar time (Figure 1B) from 7% pre-1981 to 3040% in 19862001 (
2 = 12.1; P = 0.02), again with a similar pattern for moderate-severe disease only. The prevalence among the females diagnosed since 1995 was 49/135 (36.3%; 95% CI 28.244.4%), which was not significantly different (P = 0.1) from the prevalence reported previously for the pre-1995 data (Hadfield et al., 1997
) (29.0%; 95% CI 23.734.4%). Mean age at death was lowest (17.7 years; SD 5.0) in animals that died between 1981 and 1985. The difference in diagnostic rates over the years became non-significant after adjustment for differences in age at death (P = 0.07).
|
Segregation ratios were calculated for all affected and unaffected mothers in the colony up to September 2001 (Table II). The segregation ratio for affected mothers (44.2%) was not significantly higher than for unaffected mothers (36.6%; P = 0.4); this translated into an odds ratio (OR) of endometriosis of 1.4 (95% CI 0.62.1) for daughters with affected compared with those with unaffected mothers. When the disease definition was limited to moderate-severe only (not shown), the segregation ratios were 38.7 and 29.7%, respectively (P = 0.3), corresponding to an OR of 1.5 (95% CI 0.63.5). Mothers were more likely than daughters to have died in the years up to 1985 when observed prevalence rates in the colony were reduced. Of the 101 unaffected mothers contributing to the segregation ratios, 30 (29.7%) had died pre-1986; of the 88 unaffected daughters, four (4.5%) had died pre-1986.
|
|
|
![]() |
Discussion |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Whilst there is considerable evidence that familial clustering of endometriosis occurs in humans (Zondervan et al., 2001), the present study is the first to show detailed evidence of such clustering in the rhesus macaque using pedigree information. First, the average kinship coefficient among the group of affected animals was significantly higher than among a group of randomly selected unaffecteds in the colony. Secondly, the RRs among various relative pairs of different kinship in the pedigrees generally showed a reduced risk among more distantly related animals. Thirdly, the segregation ratio among affected mothers was higher than among unaffected mothers, although this finding was not significant. It should be noted that most (80%) of the endometriosis found in this colony of rhesus macaques constituted moderate or severe disease, and that this phenotype therefore formed the main basis for our results. Separate analysis of familial aggregation limited to animals with moderate-severe disease did not materially alter the results. Unfortunately, the limited number of animals diagnosed with mild disease did not allow a separate analysis of this phenotype.
The interpretation of RR estimates in this study is complicated by potential confounding factors. First, confounding by diagnostic time-period may have occurred for the calculation of RR among relative pairs spanning different generations, such as motherdaughter or grandmothergranddaughter pairs. Analyses showed that (grand)mothers were more likely to have died prior to 1985 than their (grand)daughters, a period in which prevalence of endometriosis was lower mainly because of a lower average age at death (although the possibility of greater misclassification of endometriosis cases as unaffecteds in earlier years cannot be excluded). RRs for motherdaughter pairs (0.44; kinship coefficient 0.25) and grandmothergranddaughter pairs (0.17; kinship coefficient 0.0625) are therefore likely to be underestimates.
RRs among siblings are less likely to have been influenced by generational confounding. Moreover, any misclassification of endometriosis cases as unaffecteds within relative pairs of the same generation, if present, would have resulted in a general reduction of absolute RR values, but would not have affected the validity of comparative analyses. Although based on small numbers, we found an RR for full sibs (kinship coefficient 0.25) of 0.75, compared with RRs for paternal and maternal half sibs (kinship coefficient 0.125) of 0.50 and 0.25, respectively. Paternal half sib RR was based on much larger sibship sizes (two to 12 daughters per father) than maternal half sib risk (two to three daughters per mother). Results showed that paternal half sib risk was mainly driven by high RR among daughters from fathers with a great number of offspring. When we limited the analyses to fathers with up to three daughters, the RR decreased to 0.18, a value similar to that found for maternal half sibs. This implied that paternal half sib risk was overestimated because of the higher risk inferred by a few fathers with a large number of daughters. This could be explained by a scenario in which a few of the fathers carried a high-risk susceptibility gene, which, because they happened to produce a large number of offspring, had a large effect on RR. Interestingly, the father inferring the highest RR among paternal half sibs (an RR of 0.83 based on seven daughters, of which six were affected), also fathered one of the concordant full sib pairs. We were unable to check whether any of the fathers inferring a half sib RR >0.5 were related to each other, as they all had unknown parents in our database.
These observations highlight a further issue complicating RR interpretation in this study, namely that all relative pairs and sibships in the analyses were derived from one interconnected pedigree rather than from independent pedigrees. However, this could only have affected RR estimates if there was true genetic transmission of endometriosis in the pedigree. Suppose a susceptibility gene had been transmitted from a single founder animal to certain animals in the pedigree. All relative pairs that descended from the founder would on average display a higher RR than non-descendants. Within one pedigree, the proportion of relative pairs that happened to descend from the high-risk founder may be different for different types of relative pairs, and will influence the RR estimates. It is unlikely, however, that this bias would completely eradicate any correlation between kinship and RR. If there was no genetic transmission, we would expect RRs among the various relative pairs in the pedigree to be uncorrelated with level of kinship, provided kinship was not correlated with exposure to potential environmental risk factors. The latter assumption holds in our data, as animals are housed individually and level of kinship between animals has never been grounds for exposure to certain environmental risk factors. The general trend of higher RRs among more closely related animals in our data thus supports genetic transmission of endometriosis.
The segregation ratio among affected mothers was at 44.2% only modestly higher than that among unaffected mothers (36.6%). This could simply be a reflection of the multifactorial origin of endometriosis. In complex traits, the fact that segregation ratios do not differ between affecteds and unaffecteds does not necessarily imply a lack of genetic influence, merely a lack of a major gene effect. However, it is possible that generational confounding mentioned previously caused differential misclassification of affected mothers as unaffected, which would have reduced the difference in segregation ratios.
The prevalence of endometriosis found among all female animals in the colony that died aged 10 years was high, at 31.4% (95% CI 26.935.9%). Prevalence increased from the age of 10 years to a peak of
40% in animals aged 15 years or older at death. The analyses of prevalence by calendar year of diagnosis showed a deficit of diagnosed endometriosis cases in the years up to 1985, which was to a limited extent explained by differences in mean age at death for monkeys dying in those years. It is possible that these differences may also partly be accounted for by different frequencies of experimental procedures predisposing to endometriosis, such as hysterotomy and administration of E2. Another possibility, however, is a greater degree of misclassification of endometriosis cases as unaffecteds in earlier years, followed by an increased awareness and improved reporting of the condition in subsequent years. This is supported by an increased frequency of animals diagnosed with minimal-mild disease in later years. Similar patterns of increased prevalence of endometriosis over time (or rather, underestimated prevalence in earlier years) due to diagnostic bias has been observed in humans (Koninckx, 1998
).
We were unable to investigate the effect of (peri)menopause on the results, since we had no menstrual data on the animals. If menopausal onset is most likely to occur after the age of 25 years, then a higher probability of misclassification could have been present in the 30 animals (10%) in this age group out of the 284 unaffecteds in our analyses. Considering that such differential misclassification would be much more likely for minimal-mild disease, which has a much lower prevalence, the impact of this potential bias is likely to be small. Moreover, any such bias would have resulted in an underestimation of the prevalence in the colony, and also of the familial aggregation indicators.
We were also unable to assess exposure to environmental factors in the present study, as exposure information between 1995 and 2002 was not yet available. Using data up to 1995, we previously reported an increased risk of endometriosis with increased exposure to hysterotomy and E2; no association was found with parity or number of laparoscopies (Hadfield et al., 1997). Other risk factors that influence disease development and progression remain to be identified, but may include menstrual characteristics, differences in housing conditions and research protocol assignment. Some other environmental risk factors in the rhesus monkey that have already been identified, such as exposure to 2,3,7,8-tetrachlorodibenzo-p-dioxin (Rier et al., 1993
) and ionizing radiation (Fanton and Golden, 1991
) are unlikely to have contributed to disease development in this colony, as no animal in the pedigree has experimentally been exposed to these factors. The probability that a significant proportion of the observed familial aggregation could be explained by exposure to environmental factors remains untested, but appears unlikely because familial relationships have not been a basis for allocation of animals to research protocols, and animals are housed individually according to standard conditions.
We have described the evidence for familial aggregation of endometriosis in a well-characterized colony of rhesus macaques. Although morphologically identical to endometriosis found in women, disease found in the rhesus macaque more often represents moderate to severe stages of endometriosis. It should be borne in mind that our results of familial aggregation were predominantly based on these more severe stages of disease.
The data presented in this paper were based on the June 2002 disease status update. All 283 females still alive in the pedigrees continue to be followed up until disease status is revealed at surgery or necropsy, which will in due course enable more rigorous analyses than presented here based on the relatively small number of 142 affected animals. To investigate to what extent the observed familial aggregation can be explained by genetic susceptibility and by a particular model of genetic transmission (mode of inheritance), heritability and complex segregation analyses are needed that allow for the age-dependent risk and diagnostic cohort effects in the pedigree. Past results for other illnesses, such as breast and prostate cancer, have shown that findings from complex segregation analyses in complex traits can provide important support for gene mapping studies if evidence for a major gene effect is found (Jarvik, 2002). A genome-wide marker set specific to the rhesus macaque is currently being developed for such mapping studies (J.Roberts, personal communication). Thus, combined with the prospectively collected environmental risk factor information, the rhesus macaque data provide a unique resource to study the genetic epidemiology of endometriosis, and could provide vital new hypotheses for study of the disease in women.
![]() |
Acknowledgements |
---|
![]() |
References |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Catchpole HR and van Wagenen G (1975) Reproduction in the rhesus monkey, Macaca mulatta. In Bourne GH (ed) The Rhesus Monkey. Academic Press, San Diego, CA, pp. 117140.
Colman R and Kemnitz J (1998) Aging experiments using nonhuman primates. In Yu BP (ed) Methods in Ageing Research. CRC Press, Boca Raton, FL, pp. 249267.
Coxhead D and Thomas EJ (1993) Familial inheritance of endometriosis in a British population. A case control study. J Obstet Gynaecol 13,4244.
DHooghe TM, Bambra CS, Isahakia M and Koninckx PR (1992) Evolution of spontaneous endometriosis in the baboon (Papio anubis, Papio cynocephalus) over a 12-month period. Fertil Steril 58,409412.[ISI][Medline]
DiZerega GS, Barber DL and Hodgen GD (1980) Endometriosis: role of ovarian steroids in initiation, maintenance, and suppression. Fertil Steril 33,649653.[ISI][Medline]
Eskenazi B and Warner ML (1997) Epidemiology of endometriosis. Obstet Gynecol Clin North Am 24,235258.[ISI][Medline]
Fanton JW and Golden JG (1991) Radiation-induced endometriosis in Macaca mulatta. Radiat Res 126,141146.[ISI][Medline]
Hadfield R, Mardon H, Barlow DH and Kennedy SH (1996) Delay in diagnosis of endometriosis: a survey of women from the USA and the UK. Hum Reprod 11,878880.[Abstract]
Hadfield R, Yudkin PL, Coe CL, Scheffler J, Uno H, Barlow DH, Kemnitz JW and Kennedy SH (1997) Risk factors for endometriosis in the rhesus monkey (Macaca mulatta): a casecontrol study. Hum Reprod Update 3,109115.
Jarvik GP (2002) Complex segregation analyses: uses and limitations. Am J Hum Genet 63,942946.[CrossRef]
Kennedy SH, Bennett S and Weeks DE (2001) Genetics and infertility II: Affected sib-pair analysis in endometriosis. Hum Reprod Update 7,411418.
Koninckx PR (1998) Biases in the endometriosis literature. Illustrated by 20 years of endometriosis research in Leuven. Eur J Obstet Gynecol Reprod Biol 81,259271.[CrossRef][ISI][Medline]
Lamb K, Hoffmann RG and Nichols TR (1986) Family trait analysis: a casecontrol study of 43 women with endometriosis and their best friends. Am J Obstet Gynecol 154,601.[ISI][Medline]
MacKenzie WF and Casey HW (1975) Animal model of human disease. Endometriosis. Animal model: endometriosis in rhesus monkeys. Am J Pathol 80,341344.[ISI][Medline]
Moen MH and Magnus P (1993) The familial risk of endometriosis. Acta Obstet Gynecol Scand 72,560564.[ISI][Medline]
Olson JM and Cordell HJ (2000) Ascertainment bias in the estimation of sibling genetic risk parameters. Genet Epidemiol 18,217235.[ISI][Medline]
Rier SE, Martin DC, Bowman RE, Dmowski WP and Becker JL (1993) Endometriosis in rhesus monkeys (Macaca mulatta) following chronic exposure to 2,3,7,8-tetrachlorodibenzo-p-dioxin. Fundam Appl Toxicol 21,433441.[CrossRef][ISI][Medline]
Schenken RS, Williams RF and Hodgen GD (1987) Effect of pregnancy on surgically induced endometriosis in cynomolgus monkeys. Am J Obstet Gynecol 157,13921396.[ISI][Medline]
Simpson JL, Elias S, Malinak LR and Buttram VCJ (1980) Heritable aspects of endometriosis. I. Genetic studies. Am J Obstet Gynecol 137,327331.[ISI][Medline]
Stefansson H, Geirsson RT, Steinthorsdottir V, Jonsson H, Manolescu A, Kong A, Ingadottir G, Gulcher J and Stefansson K (2002) Genetic factors contribute to the risk of developing endometriosis. Hum Reprod 17,555559.
The American Fertility Society (1985) Revised American Fertility Society classification of endometriosis: 1985. Fertil Steril 43,351352.[Medline]
Treloar SA and KennedySH (2002) Preliminary results from two combined genome-wide scans in endometriosis. Fertil Steril 77,S19S20.
Treloar SA, OConnor DT, OConnor VM and Martin NG (1999) Genetic influences of endometriosis in an Australian twin sample. Fertil Steril 71,701710.[CrossRef][ISI][Medline]
VandeBerg JL and Williams-Blangero S (1997) Advantages and limitations of nonhuman primates as animal models in genetic research on complex diseases. J Med Primatol 26,113119.[ISI][Medline]
Zondervan KT, Cardon LR and Kennedy SH (2001) The genetic basis of endometriosis. Curr Opin Obstet Gynecol 13,309314.[CrossRef][ISI][Medline]
Zondervan KT, Cardon LR and Kennedy SH (2002) What makes a good casecontrol study? Design issues for complex traits such as endometriosis. Hum Reprod 17,14151423.
Submitted on May 27, 2003; accepted on October 2, 2003.