Prevalence and risk factors of adenomyosis at hysterectomy

T. Bergholt1,3, L. Eriksen1, N. Berendt2, M. Jacobsen2 and J.B. Hertz1

1 Department of Obstetrics and Gynaecology and 2Department of Pathology, Gentofte University Hospital, Niels Andersens Vej 65, 2 900 Hellerup, Denmark


    Abstract
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
BACKGROUND: The present study was performed to evaluate the prevalence and possible associated risk factors for adenomyosis. METHODS: Medical records were retrieved and histo-pathological material re-examined for 549 consecutive women undergoing hysterectomy in a two-year period from 1990–1991. RESULTS: The prevalence of adenomyosis in the study varied from 10.0–18.2%, depending on different diagnostic criteria. The presence of endometrial hyperplasia at the time of hysterectomy was the only variable significantly associated with adenomyosis (OR = 3.0; 95% CI: 1.2–8.3). No statistically significant association was found between adenomyosis and previous caesarean section, endometrial curettage or evacuation of the uterus. Furthermore, we did not see any significant association between adenomyosis and pain-related symptoms, indication for hysterectomy, age, parity or number of myometrial samples. CONCLUSIONS: Our study stresses the need for precise diagnostic criteria for adenomyosis, and furthermore indicates that endometrial hyperplasia and adenomyosis may have a common aetiology.

Key words: adenomyosis/prevalence/risk factors


    Introduction
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Adenomyosis uteri is a pathological entity characterized by the presence of endometrial glands and stroma embedded within the myometrium without apparent contact with the endo-myometrial junction. As the diagnosis of adenomyosis is based on histological examination, the condition is best described in women at the time of hysterectomy. Adenomyosis is often seen in peri-menopausal women, and is suggested to be related to bleeding disorders, dysmenorrhoea and parity. The prevalence has been reported to range from 8.8–61.5% in women at the time of hysterectomy (Hunter et al., 1947Go; Israel and Wountersz, 1959; Molitor, 1971Go; Bird et al., 1972Go; Olawabi and Strickler, 1977Go; Lee et al., 1984Go; Shaikh and Khan, 1990Go; Seidman and Kjerulff, 1996Go), and some authors have described this entity as elusive (Bird et al., 1972Go). As the results are conflicting, several potential forms of bias may have influenced the opposing findings of these studies. Various reports have outlined the appearance of symptomatic endometrial tissue in the abdominal wall following amniocentesis and pelvic surgery, especially following caesarean section (Koger et al., 1993Go; Hughes et al., 1997Go; Liang et al., 1998Go). This phenomenon has been explained by implantation during surgery, followed by the embedding and survival of the ectopic endometrium. For this reason, a history of transuterine and transcervical surgery, and other potential variables, were introduced into the present study in order to determine whether they could be risk factors for adenomyosis.


    Materials and methods
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
In the period from January 1990 through to December 1991, 549 consecutive patients underwent hysterectomy at the Department of Obstetrics and Gynaecology, Gentofte University Hospital. The hospital is a tertiary teaching hospital, serving a population of ~200 000 in Copenhagen County. The medical records were reviewed by the authors, and age, parity, history of previous caesarean section, curettage or evacuation of the uterus, pre-operative dyspareunia, dysmenorrhoea or chronic pelvic pain, indications for hysterectomy, and post-operative findings, including the presence of peritoneal endometriosis, were obtained, together with the original pathological reports. Caesarean section, uterine evacuation and transcervical curettage were recorded in this study only if performed more than 6 months previous to hysterectomy. This was done in order to avoid temporarily embedded endometrial tissue inside the myometrium being diagnosed as adenomyosis.

All the histological specimens were re-examined by the same pathologists, with special emphasis on the appearance of adenomyosis, the distance between foci of adenomyosis and the endo-myometrial junction, and the presence of surrounding myometrial hyperplasia. Other histopathological variables were also noted, such as endometrial status and the number of histological samples from the endo-myometrium of each uterus.

A total of 538 women (98%) were scheduled for abdominal hysterectomy, and 11 (2%) for vaginal hysterectomy. The mean age was 54 years (range 23–88). Indications were divided into four main categories, with 183 women having more than one indication. Bleeding disorders were found in 278 women (50.6%), malignant neoplasia in the genital tract in 185 (33.7%), pelvic pain in 146 (26.6%) and pelvic relaxation in 142 women (25.9%). The most frequent combination of diagnosis was bleeding disorder and pelvic pain, in 123 cases. Only 23 women underwent hysterectomy with pelvic pain as the only indication.

In the statistical analysis, the odds ratios (OR) and their corresponding 95% confidence interval (CI) were calculated using multiple logistic regression analysis in SAS for Windows, Release 6.12. If the value 1.0 was not covered by the CI, the test result was considered statistically significant (P < 0.05). The study was approved by the Local Ethical Committee of Copenhagen County.


    Results
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
The prevalence of adenomyosis in our study is presented in Table IGo according to (i) the various distances between the endomyometrial junction and the endometrial glands and stroma and (ii) the presence or absence of myometrial hyperplasia. As may be seen, the prevalence ranges from 10.0–18.2%. In the subsequent analysis, we decided on the following criterion for adenomyosis: >=3 mm distance from the foci of the glands and stroma to the endo-myometrial junction and surrounding myometrial hyperplasia, resulting in a prevalence of 12.5%. Of these 68 cases, 45 were histologically classified as diffuse, and the rest as focal.


View this table:
[in this window]
[in a new window]
 
Table I. Diagnostic criteria and prevalence of adenomyosis in 549 women undergoing hysterectomy in Gentofte University Hospital, 1990–1991
 
A history of previous Caesarean section, endometrial curettage or evacuation did not demonstrate any association with adenomyosis in the present study population, as is presented in Table IIGo. Moreover, no association was established between adenomyosis and parity or the number of histological samples from the endo-myometrium. However, an age of >54 years had an adjusted OR = 2.4 (95% CI: 1.0–5.8), indicating a tendency towards an association with adenomyosis. Table IIIGo presents the association between pain-related symptoms and adenomyosis. No relationship was found between dyspareunia, dysmenorrhea, chronic pelvic pain and adenomyosis. With regard to the four categories of indications used in this study, none demonstrated significant associations with the presence of adenomyosis in the adjusted multivariate analysis, as presented in Table IVGo. In the analysis of the endometrial status demonstrated in Table VGo, only hyperplasia (OR = 3.0; 95% CI: 1.2–8.3) demonstrates a significant positive association with adenomyosis.


View this table:
[in this window]
[in a new window]
 
Table II. Previous uterine surgery, parity, age and number of endo-myometrial samples and crude/adjusted odds ratios for the prevalence of adenomyosis at hysterectomy
 

View this table:
[in this window]
[in a new window]
 
Table III. Pelvic pain symptoms and the corresponding crude/adjusted odds ratios for the prevalence of adenomyosis at hysterectomy
 

View this table:
[in this window]
[in a new window]
 
Table IV. Indications and the corresponding crude/adjusted odds ratios for the prevalence of adenomyosis at hysterectomy. Each patient may have more than one indication
 

View this table:
[in this window]
[in a new window]
 
Table V. Endometrial status and the corresponding crude/adjusted odds ratios for the prevalence of adenomyosis at hysterectomy
 

    Discussion
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
In the study of adenomyosis, the optimal study-population would consist of women with a uterus in situ. As the diagnosis of this condition at the present time is based on histological examination of the entire uterus, present knowledge is mainly based on women undergoing hysterectomy. As there is great variation in the use and indication for hysterectomy on national as well as international levels (Haas et al., 1993Go; Hall and Cohen, 1994Go, Keskimaki et al., 1994Go; Ferenczy, 1998Go) populations of women undergoing hysterectomy in different regions are heterogeneous, and consequently subject to demographic variation. This fact is probably responsible for some of the great variation seen in the prevalence of adenomyosis at hysterectomy worldwide.

The present study has demonstrated that the prevalence of adenomyosis at hysterectomy varies from 10.0–18.2%, depending on different histological criteria with regard to the distance from the endo-myometrial junction to the foci of adenomyosis and the presence or absence of myometrial hyperplasia around the foci. Most previously published studies utilize ‘one low-power field’ as the minimal distance in the diagnostic criteria. Great variation has been demonstrated in the area of microscopic fields using different microscopes, and the authors conclude that the exact measure of the area should be used (Ellis and Whitehead, 1981Go). This problem is also discussed by Ferenczy, suggesting a relative measure of 25% of the uterine wall as the ideal distance for the diagnosis of adenomyosis (Ferenczy, 1998Go). Inaccurate measuring could explain some of the great variation in the prevalence of adenomyosis. A third factor could be the number of histological samples taken for investigation. A proportional correlation between the number of samples taken and the presence of adenomyosis in the same uterus has been shown (Bird et al., 1972Go). In addition, it has been suggested that the pathologists' awareness of this condition could have an influence on the prevalence. In order to eliminate this possible information bias, we decided to re-examine all the histological specimens using strict objective criteria. We suggest the criteria for adenomyosis should be the presence of endometrial foci embedded in the myometrium at least 3 mm from the endo-myometrial junction surrounded by myometrial hyperplasia. This includes an absolute distance and consequently should be easy to reproduce, irrespective of the microscope used.

In recently published studies, parity, previous spontaneous abortions, dilatation and curettage for gynaecological indications, age between 40 and 59 years, and endometrial hyperplasia have been positively statistically associated with adenomyosis, whereas smoking was found to be negatively correlated (Shaikh and Khan., 1990Go; Vercellini et al., 1995Go; Parazzini et al., 1997Go). In the present study we have introduced all of these possible confounding variables into our multiple regression model, with the exception of data on smoking habits, which were not available. Data retrieved from the medical records were retrospective. As they were obtained and described before the time of histological examination of the uterus, they may have lacked precision, but could not introduce bias into the statistical analysis due to differential misclassification.

We did not find any statistical association between adenomyosis and previous transcervical or transmural surgery, and we have consequently been unable to verify the suggested hypothesis of surgically-induced endometrial implantation as a possible cause of adenomyosis. Neither did we find any significant relationship between adenomyosis and pain-related symptoms, indication for hysterectomy, age, parity or the number of endo-myometrial samples, although a slight tendency was seen in women with pre-operative dysmenorrhoea or aged 55 years or more. The only significant association in our data was the relationship between adenomyosis and endometrial hyperplasia. The result substantiates previously published findings of a relationship between high concentrations of oestrogen and adenomyosis (Parazzini et al., 1997Go), and the elevated oestrogen concentration in the menstrual blood of women with adenomyosis (Takahashi et al., 1989Go). In addition, relatively high oestrogen biosynthesis due to aromatase activity in human adenomyotic tissue has been described, and it is suggested that this contributes to the growth of adenomyosis. This synthesis of oestrogens could be blocked by danazol (Urabe et al., 1989Go; Yamamoto et al., 1993Go). Furthermore, Tamoxifen has been described as inducing adenomyosis in post-menopausal women treated for breast cancer, because of its oestrogen–agonistic effect (Cohen et al., 1997Go). Taken together, these findings indicate that oestrogen plays an essential role in the aetiology of adenomyosis. However, whether adenomyosis is the cause or effect of a local increased oestrogen concentration in the uterine environment cannot be inferred using a prevalence study design. With the continuous advance in the fields of endovaginal ultrasound and magnetic resonance scanning (Mark et al., 1987Go; Ascher et al., 1994Go; Vercellini et al., 1998Go), it may be possible in the future to perform non-invasive population-based prospective incidence studies of adenomyosis, thus throwing a clearer and more homogeneous light on the effect of oestrogen and other potential risk factors on this elusive condition.


    Notes
 
3 To whom correspondence should be addressed. E-mail: tbe{at}dadlnet.dk Back


    References
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Ascher, S.M., Arnold, L.L., Patt, R.H. et al. (1994) Adenomyosis: prospective comparison of MR and transvaginal sonography. Radiology, 190, 803–806.[Abstract]

Bird, C.C., McElin, T.W. and Manalo-Estrella, P. (1972) The elusive adenomyosis of the uterus. Am. J. Obstet. Gynecol., 112, 583–593.[ISI][Medline]

Cohen, I., Beyth, Y., Shapira, J. et al. (1997) High frequency of adenomyosis in postmenopausal breast cancer patients treated with tamoxifen. Gynecol. Obstet. Invest., 44, 200–205.[ISI][Medline]

Ellis, P.S.J. and Whitehead, R. (1981) Mitosis counting – a need for reappraisal. Hum. Pathol., 12, 3–4.[ISI][Medline]

Ferenczy, A. (1998) Pathophysiology of adenomyosis. Hum. Reprod. Update, 4, 312–322.[Abstract/Free Full Text]

Haas, S., Acker, D., Donahue, C. et al. (1993) Variation in hysterectomy rates across small geographic areas of Massachusetts. Am. J. Obstet. Gynecol., 169, 150–154.[ISI][Medline]

Hall, R.E. and Cohen, M.M. (1994) Variation in hysterectomy rates in Ontario: does the indication matter. C.M.A.J., 151, 1713–1719.[Abstract]

Hughes, M.L., Bartholomew, D. and Palluzi, M. (1997) Abdominal wall endometriosis after amniocentesis. A case report. J. Reprod. Med., 42, 597–599.[ISI][Medline]

Hunter, W.C., Smith, L.L. and Reiner, W.C. (1947) Uterine adenomyosis. Incidence, symptoms and pathology in 1856 hysterectomies. Am. J. Obstet. Gynecol., 53, 663–668.[ISI]

Israel, S.L. and Woutersz, T.B. (1959) Adenomyosis: a neglected diagnosis. Obstet. Gynecol., 14, 168–173.[ISI][Medline]

Keskimaki, I., Aro, S. and Teperi, J. (1994) Regional variation in surgical procedures in Finland. Scand. J. Soc. Med., 22, 132–138.[ISI][Medline]

Koger, K.E., Shatney, C.H., Hogde, K. et al. (1993) Surgical scar endometriosis. Surg. Gynecol. Obstet., 177, 243–246.[ISI][Medline]

Lee, N.C., Dicker, R.C., Rubin, G.L. et al. (1984) Confirmation of the preoperative diagnoses for hysterectomy. Am. J. Obstet. Gynecol., 150, 283–287.[ISI][Medline]

Liang, C.C., Liou, B., Tsai, C.C. et al. (1998) Scar endometriosis. Int. Surg., 83, 69–71.[ISI][Medline]

Mark, A.S., Hricak, H., Heinrichs, L.W. et al. (1987) Adenomyosis and leiomyoma: Differential diagnosis with MRI. Radiology, 163, 527–529.[Abstract]

Molitor, J.J. (1971) Adenomyosis: a clinical and pathological appraisal. Am. J. Obstet. Gynecol., 110, 275–284.[ISI][Medline]

Olawabi, T.O. and Strickler, R.C. (1977) Adenomyosis a neglected diagnosis. Obstet. Gynecol., 50, 424–427.[Abstract]

Parazzini, F., Vercellini, P., Panazza, L.C. et al. (1997) Risk factors for adenomyosis. Hum. Reprod., 12, 1275–1279.[ISI][Medline]

Seidman, J.D. and Kjerulff, K.H. (1996) Pathologic findings from the Maryland Women's Health Study: practice patterns in the diagnosis of adenomyosis. Int. J. Gynecol. Pathol., 15, 217–221.[ISI][Medline]

Shaikh, H. and Khan, K.S. (1990) Adenomyosis in pakistani women: four year experience at the Aga Khan University Medical Centre, Karachi. J. Clin. Pathol., 43, 817–819.[Abstract]

Takahashi, K., Nagata, H. and Kitao, M. (1989) Clinical usefulness of determination of estradiol level in the menstrual blood for patients with endometriosis. Acta Obstet. Gynaecol. Jpn., 41, 1849–1850.

Urabe, M., Yamamoto, T., Kitawaki, J. et al. (1989) Estrogen biosynthesis in human uterine adenomyosis. Acta Endocrinol. (Copenhagen), 121, 259–264.[ISI][Medline]

Vercellini, P., Parazzini, F., Oldani, S. et al. (1995) Adenomyosis at hysterectomy:A study on frequency distribution and patient characteristics. Hum. Reprod., 10, 1160–1162.[Abstract]

Vercellini, P., Cortesi, I., De Giorgi, O. et al. (1998) Transvaginal ultrasonography versus uterine needle biopsy in the diagnosis of diffuse adenomyosis. Hum. Reprod., 13, 2884–2887.[Abstract/Free Full Text]

Yamamoto, T., Noguchi, T., Tamura, T. et al. (1993) Evidence for estrogen synthesis in adenomyotic tissue. Am. J. Obstet. Gynecol., 169, 734–738.[ISI][Medline]

Submitted on April 4, 2001; accepted on July 25, 2001.