Skip Navigation


American Journal of Epidemiology Advance Access originally published online on August 5, 2008
American Journal of Epidemiology 2008 168(6):571-574; doi:10.1093/aje/kwn165
This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
168/6/571    most recent
kwn165v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Related articles in Am. J. Epidemiol.
Right arrow Similar articles in this journal
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrowRequest Permissions
Right arrow Disclaimer
Google Scholar
Right arrow Articles by Lacey, J. V.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Lacey, J. V., Jr.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

American Journal of Epidemiology Published by the Johns Hopkins Bloomberg School of Public Health 2008.

Invited Commentary

Invited Commentary: Endometrial Hyperplasia—Getting Back to Normal

James V. Lacey, Jr.

From the Hormonal and Reproductive Epidemiology Branch, Division of Cancer Epidemiology and Genetics, National Cancer Institute, Rockville, MD

Correspondence to Dr. James V. Lacey, Jr., Division of Cancer Epidemiology and Genetics, National Cancer Institute, 6120 Executive Blvd., MSC 7234, Rockville, MD 20852-7234 (e-mail: jimlacey{at}nih.gov).

Received for publication March 25, 2008. Accepted for publication April 24, 2008.


   ABSTRACT
TOP
 ABSTRACT
References
 
Cancer precursors can help to reveal clues about how and when risk factors influence the development of carcinoma. Endometrial carcinoma is well-suited to studies of precursors: Strong risk and protective factors exist, as does a good candidate precursor lesion, called atypical endometrial hyperplasia. Atypical hyperplasia is the most severe type of endometrial hyperplasia, which ranges from mild, reversible proliferation to incipient carcinoma. In this issue of the Journal, Epplein et al. (Am J Epidemiol 2008;168:563–70) report that three established risk factors for endometrial carcinoma—obesity, parity, and smoking—are similarly associated with two types of endometrial hyperplasia: complex hyperplasia and atypical hyperplasia. How much these findings reveal about mechanistic pathways for endometrial carcinoma depends, in part, on three issues that specifically affect endometrial hyperplasia but also affect other precursors. They are: 1) potential misclassification of intermediate endpoints, 2) unsettled thresholds between low-risk and high-risk lesions, and 3) uncertain boundaries between normal tissue and early-stage precursors. In this commentary, the author explores how these issues might influence interpretation of the new data from Epplein et al. and shape future research on endometrial carcinoma precursors.

endometrial hyperplasia; endometrial neoplasms; endometrium; obesity; parity; smoking

Abbreviations: WHO, World Health Organization

Cancer precursors facilitate early detection, serve as efficient surrogate endpoints for carcinoma in large research studies, and provide targets for chemoprevention. Endometrial carcinoma has a well-recognized precursor, called atypical hyperplasia. Atypical hyperplasia is the most severe form of endometrial hyperplasia. The diagnosis of endometrial hyperplasia covers a broad spectrum of lesions, ranging from mild proliferation that spontaneously regresses to highly irregular glands and nuclei that merge imperceptibly with endometrial carcinoma. The current version of the World Health Organization (WHO) classification system for endometrial hyperplasia uses two criteria—glandular architecture (simple vs. complex crowding) and nuclear atypia (nonatypical vs. atypical)—to classify endometrial hyperplasia as simple, complex, or atypical (1). These classifications guide clinical management of women who receive a diagnosis of endometrial hyperplasia, typically based on endometrial biopsy or curettage (2), even though the WHO system's limitations have prompted discussion of ways to improve or replace it (3). Despite endometrial carcinoma's attractiveness as a research topic—strong risk factors (e.g., obesity and unopposed estrogen therapy) and protective factors (parity and oral contraceptive use) exist, it is a common gynecologic tumor, and hysterectomy and outpatient endometrial sampling provide accessible tissue specimens for research—studies of endometrial hyperplasia as a precursor for endometrial carcinoma have been surprisingly sparse.

This issue of the Journal includes a paper by Epplein et al. (4) that helps to fill this gap. They investigated whether hormonal risk factors for endometrial carcinoma are also associated with the risk of developing complex hyperplasia and atypical hyperplasia. Using a case-control design, they reviewed original slides from 1,784 members of Group Health (Seattle/Puget Sound, Washington) who had been diagnosed with complex hyperplasia or atypical hyperplasia between 1985 and 2003 to identify 271 cases with complex hyperplasia and 175 cases with atypical hyperplasia. Risk-set sampling yielded appropriately matched controls from the same health plan. Medical records provided risk factor data, with essentially complete retrieval of risk factor data for cases and controls. Studies like this highlight the advantages of conducting research in large managed health care plans.

Epplein et al. (4) concluded that obesity, parity, and smoking are associated with complex hyperplasia and atypical hyperplasia, just as they are with endometrial carcinoma. The odds ratios and 95 percent confidence intervals for obesity and morbid obesity were imprecise but were consistent with previously reported associations between endometrial carcinoma and a body mass index (weight (kg)/height (m)2) over 25. Multiparity was strongly inversely associated with complex hyperplasia and atypical hyperplasia, although the odds ratios were slightly attenuated among women aged 52 years or more. Current smoking was inversely, though not statistically significantly, associated with atypical hyperplasia. With similar risk factors for endometrial hyperplasia and carcinoma, Epplein et al. close their article with a logical question: Are these risk factors for initiation of complex hyperplasia and atypical hyperplasia or for progression from complex hyperplasia and atypical hyperplasia to carcinoma? Answering that question depends on how we address three major challenges of studying endometrial hyperplasia: misclassification, progression risks, and comparison groups.

Let's start with misclassification. Each of the widely used WHO classifications (simple hyperplasia, complex hyperplasia, and atypical hyperplasia) has low reproducibility (3). Simple hyperplasia is often overdiagnosed. Upon rereview, many lesions originally classified as simple hyperplasia are downgraded to something less than endometrial hyperplasia—normal proliferative, secretory, or atrophic endometrial tissue. In contrast, atypical hyperplasia is often an underdiagnosis of endometrial carcinoma. Up to 50 percent of women who receive an endometrial-biopsy-based diagnosis of atypical hyperplasia are shown to have endometrial carcinoma when hysterectomy is performed weeks or months later (5–8). Endometrial biopsies capture only a sample of endometrium. If the sampling misses the sometimes-small loci of severe endometrial hyperplasia or carcinoma, the resulting endometrial hyperplasia diagnosis will underestimate the lesion's true severity. When combined with pathologists' varying thresholds for deciding between atypical hyperplasia and carcinoma, it can be difficult to determine whether an endometrial biopsy classified as endometrial hyperplasia has been overcalled, undercalled, or correctly classified.

Epplein et al. took the best possible approach to the problem. They enlisted experienced pathologists to review original slides and assign study diagnoses. Not surprisingly, they downgraded two thirds of the cases' original complex hyperplasia and atypical hyperplasia diagnoses to simple hyperplasia or normal. Most cancer case-control studies have highly reproducible case groups. In contrast, the threshold for ineligible simple hyperplasia versus eligible complex hyperplasia in Epplein et al.'s study will be difficult to replicate, because the criteria for differentiating simple hyperplasia from complex hyperplasia are subjective and imprecise. This reflects suboptimal WHO criteria (and holds true for other intra-endometrial hyperplasia comparisons, too), not poor study design or data collection.

Given the aim of the Epplein et al. study—to study risk factors for endometrial hyperplasia, not cancer—misclassification at the other end of the spectrum is more worrisome. The authors excluded any case whose biopsy was called carcinoma by community or research pathologists. Subsequent analyses that excluded 41 cases who underwent a hysterectomy that revealed carcinoma within 8 weeks of their complex hyperplasia or atypical hyperplasia diagnosis reportedly produced associations similar to those of analyses using all cases (4). The sensitivity of this 8-week criterion for identifying all occult carcinomas is unknown but might be too low. In progression studies (9, 10), investigators have considered any carcinoma diagnosed within 1 year of the endometrial hyperplasia to reflect prevalent carcinoma and have excluded those patients from study. Most tumors develop over years or decades, not weeks or months, so even a 1-year time lag might be insufficient.

Epplein et al. noted only the possibility of misclassified endpoints. It is fairly straightforward to predict the direction of the effect of bias due to occult carcinomas' being misclassified as atypical hyperplasia cases. Risk factor associations for atypical hyperplasia will look too similar to risk factor associations for endometrial carcinoma when an atypical hyperplasia case group includes patients with carcinoma. Estimating the magnitude of the potential bias is more difficult. The observed odds ratios for obesity and atypical hyperplasia were 3.7 (95 percent confidence interval: 1.0, 13.8) among younger women (age <52 years) and 1.6 (95 percent confidence interval: 0.8, 3.1) among older women (age ≥52 years). Epplein et al. reported that 23 percent of the atypical hyperplasia cases (41 of 175) were diagnosed with carcinoma within 8 weeks of their atypical hyperplasia diagnosis. Other studies suggest that the true prevalence of concurrent carcinoma among patients with atypical hyperplasia is twice as high (11). If up to one half of the atypical hyperplasia cases—not just the 41 cases excluded in sensitivity analyses—had occult carcinoma, then the true association between obesity and atypical hyperplasia only (i.e., after excluding all of those patients) might be lower, and perhaps even indistinguishable from the null. The reported odds ratios for morbid obesity (i.e., body mass index ≥40) were higher but were based on very few controls, and the inverse association between current smoking and atypical hyperplasia was also based on small sample sizes. Here the strong associations between endometrial carcinoma and obesity and smoking complicate the interpretation of associations between atypical hyperplasia and these risk factors: Even slight misclassification of atypical hyperplasia cases could generate positively biased associations with atypical hyperplasia. More extensive sensitivity analyses from Epplein et al.—for example, what was the association among the excluded patients, rather than among the large group of remaining cases and controls?—could have addressed this concern.

The boundary between atypical hyperplasia and carcinoma brings us to the issue of progression risks. At present, atypical hyperplasia appears to be the single high-risk type of endometrial hyperplasia. Evidence-based treatment guidelines do not exist, but current opinion favors hysterectomy for patients with atypical hyperplasia or carcinoma (12, 13). In our recent study of endometrial hyperplasia progression, persons with atypical hyperplasia had a significantly elevated short-term and long-term progression risk, whereas persons with nonatypical endometrial hyperplasia (i.e., simple hyperplasia or complex hyperplasia) were at low risk and unlikely to progress to carcinoma, given contemporary progestin-based treatment and clinical follow-up (9). In fact, the progression risk for complex hyperplasia was not markedly higher than the progression risks for simple hyperplasia and disordered proliferative endometrium, which is a lesion less severe than endometrial hyperplasia (5, 14). By considering both atypical hyperplasia and complex hyperplasia to be high-risk and excluding simple hyperplasia from their study, Epplein et al. indirectly emphasized complex hyperplasia. How one views complex hyperplasia influences how one develops testable hypotheses about initiation versus progression. Is complex hyperplasia merely a more worrisome-looking low-risk lesion that, like simple hyperplasia, will probably regress, or is it almost a true precursor, like atypical hyperplasia? If, as Epplein et al. suggest, complex hyperplasia is high-risk and differs from simple hyperplasia, then investigators in future studies should search for factors that differentiate simple hyperplasia from complex hyperplasia. If, instead, complex hyperplasia is low-risk, like simple hyperplasia—and our progression data support that view (9)—then Epplein et al.'s analysis implies that obesity and nulliparity (and perhaps never smoking) are risk factors for all endometrial hyperplasia because they were similarly associated with both low-risk and high-risk types. This also suggests that other genetic, molecular, or clinical characteristics then determine which endometrial hyperplasia lesions progress to carcinoma.

The third issue affects progression studies in general but is particularly important for endometrial carcinogenesis: When a case group is defined as being at an intermediate step on the pathway to carcinoma, how "normal" should the reference group be? Figure 1 shows a conceptual view of the progression from normal endometrium to endometrial hyperplasia to carcinoma, plus selected characteristics of those stages. Viewing the cases and controls from the Epplein et al. study within this framework helps us to understand what the reported associations say about where and when these risk factors might operate. Epplein et al. excluded all cases originally diagnosed with carcinoma, called everyone with complex hyperplasia or atypical hyperplasia a case, and excluded women with simple hyperplasia, leaving women with presumably normal endometrial tissue as controls. Therefore, their analysis compared women in the middle-right part of this progression with women at baseline. By excluding simple hyperplasia patients, rather than including them as another case group or as controls, the authors missed the chance to explore the potentially critical transitions from normal endometrium to simple hyperplasia and from simple hyperplasia to complex hyperplasia.


Figure 1
View larger version (13K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		FIGURE 1. Conceptual model of the natural history of endometrial carcinoma. For state, the thickness of the lines denotes the relative balance of progression (forward arrows) and regression (backward arrows). The quantitative data and treatment guidelines presented are based on the current literature (1–3, 9, 14).

 
Uncertainty about whether clinical sampling identifies all carcinomas, what differentiates low-risk lesions from high-risk lesions, and how low-risk a referent group should be also affects precursors for other kinds of cancers, including thyroid (15), breast (16, 17), and colorectal (18, 19) cancers, to name just a few. Multiple approaches can be used tackle these problems, and Epplein et al. offered reasonable arguments for their design, analysis, and interpretation. On three other issues, additional data from Epplein et al. would have helped readers interpret the results. First, instead of determining precise menopausal status for cases at the time of their endometrial hyperplasia diagnosis, the authors used age (52 years) as a surrogate. This limits generalizability, because menopausal status influences whether endometrial hyperplasia is treated hormonally (for premenopausal women) or surgically (for postmenopausal women). The traditional premenopausal-versus-postmenopausal dichotomy used in most epidemiologic studies ignores the true complexity of the multistage menopausal transition (20), especially for women with endometrial hyperplasia, who are probably somewhere in between. There is an acute need, both methodologically and clinically, to better understand this transition. Second, Epplein et al. seemed to imply that abnormal uterine bleeding is on the causal pathway to endometrial hyperplasia and carcinoma. Abnormal bleeding is the presenting symptom for almost all instances of endometrial hyperplasia and carcinoma, but only one quarter of women who undergo endometrial biopsy for abnormal uterine bleeding have endometrial hyperplasia or carcinoma (12). Atrophy, endometritis, and polyps account for most instances of abnormal bleeding (21, 22). Some common etiology with endometrial hyperplasia is expected but has not been definitively proven. In future studies of endometrial hyperplasia, investigators may wish to use abnormal uterine bleeding to screen large clinical populations, enrich study populations for endometrial hyperplasia and carcinoma, and capture the full range of "normal" controls. Third, oral contraceptives and menopausal estrogen plus progestin therapy were not included. Both are essential factors for understanding endometrial carcinogenesis, not only as established risk factors (23) but also because they can both cause and treat abnormal uterine bleeding (22). Perhaps in subsequent papers Epplein et al. will fully address these risk factors in light of potential confounding by indication and misclassification of treatment versus other uses.

For endometrial hyperplasia, each potential step in the natural history represents a nondiscrete boundary where important error, bias, and uncertainty can appear. These may be prices to pay for the many advantages of having good candidate precursor lesions. With so few rigorous, population-based studies about endometrial hyperplasia risk factors and progression and the substantial questions that remain, no individual study can be expected to singly or comprehensively address the entire transition from normal status to carcinoma. Epplein et al.'s study (4) improves the literature because its sound design, high-quality data, and provocative findings identify specific, testable hypotheses for future work.


   ACKNOWLEDGMENTS
 
The author thanks Douglas A. Richesson for technical assistance.

Conflict of interest: none declared.


   References
TOP
 ABSTRACT
 References
 

  1. Silverberg SG, Mutter GL, Kurman RJ, et al. Tumors of the uterine corpus: epithelial tumors and related lesions. In: WHO classification of tumors: pathology and genetics of tumors of the breast and female genital organs—Tavassoli FA, Stratton MR, eds. (2003) Lyon, France: IARC Press. 221–32.
  2. Marsden DE, Hacker NF. Optimal management of endometrial hyperplasia. Best Pract Res Clin Obstet Gynaecol (2001) 15:393–405.[CrossRef][Medline]
  3. Zaino RJ. Endometrial hyperplasia: is it time for a quantum leap to a new classification? Int J Gynecol Pathol (2000) 19:314–21.[Web of Science][Medline]
  4. Epplein M, Reed SD, Voight LF, et al. Risk of complex and atypical endometrial hyperplasia in relation to anthropometric measures and reproductive history. Am J Epidemiol (2008) 168:563–70.[Abstract/Free Full Text]
  5. Silverberg SG. Problems in the differential diagnosis of endometrial hyperplasia and carcinoma. Mod Pathol (2000) 13:309–27.[CrossRef][Web of Science][Medline]
  6. Trimble CL, Kauderer J, Zaino R, et al. Concurrent endometrial carcinoma in women with a biopsy diagnosis of atypical endometrial hyperplasia: a Gynecologic Oncology Group study. Cancer (2006) 106:812–19.[CrossRef][Medline]
  7. Shutter J, Wright TC Jr. Prevalence of underlying adenocarcinoma in women with atypical endometrial hyperplasia. Int J Gynecol Pathol (2005) 24:313–18.[CrossRef][Medline]
  8. Valenzuela P, Sanz JM, Keller J. Atypical endometrial hyperplasia: grounds for possible misdiagnosis of endometrial adenocarcinoma. Gynecol Obstet Invest (2003) 56:163–7.[CrossRef][Web of Science][Medline]
  9. Lacey JV Jr, Ioffe OB, Ronnett BM, et al. Endometrial carcinoma risk among women diagnosed with endometrial hyperplasia: the 34-year experience in a large health plan. Br J Cancer (2008) 98:45–53.[CrossRef][Web of Science][Medline]
  10. Kurman RJ, Kaminski PF, Norris HJ. The behavior of endometrial hyperplasia. A long-term study of "untreated" hyperplasia in 170 patients. Cancer (1985) 56:403–12.[CrossRef][Web of Science][Medline]
  11. Zaino RJ, Kauderer J, Trimble CL, et al. Reproducibility of the diagnosis of atypical endometrial hyperplasia: a Gynecologic Oncology Group study. Cancer (2006) 106:804–11.[CrossRef][Medline]
  12. Montgomery BE, Daum GS, Dunton CJ. Endometrial hyperplasia: a review. Obstet Gynecol Surv (2004) 59:368–78.[CrossRef][Web of Science][Medline]
  13. Clark TJ, Neelakantan D, Gupta JK. The management of endometrial hyperplasia: an evaluation of current practice. Eur J Obstet Gynecol Reprod Biol (2006) 125:259–64.[CrossRef][Web of Science][Medline]
  14. Mazur MT. Endometrial hyperplasia/adenocarcinoma: a conventional approach. Ann Diagn Pathol (2005) 9:174–81.[CrossRef][Medline]
  15. Sarlis NJ, Gourgiotis L, Filie AC. Misclassification of cytologic diagnoses in patients with follicular lesions or follicular neoplasms of the thyroid gland: implications for patient care and clinical research. Cancer (2002) 96:323–4.[Medline]
  16. Hartmann LC, Sellers TA, Frost MH, et al. Benign breast disease and the risk of breast cancer. N Engl J Med (2005) 353:229–37.[Abstract/Free Full Text]
  17. Collins LC, Baer HJ, Tamimi RM, et al. Magnitude and laterality of breast cancer risk according to histologic type of atypical hyperplasia: results from the Nurses' Health Study. Cancer (2007) 109:180–7.[Medline]
  18. Poole C. Controls who experienced hypothetical causal intermediates should not be excluded from case-control studies. Am J Epidemiol (1999) 150:547–51.[Abstract/Free Full Text]
  19. Potter JD. Invited commentary: old problem, new wrinkles. Am J Epidemiol (1998) 147:911–13.[Free Full Text]
  20. Soules MR. Development of a staging system for the menopause transition: a work in progress. Menopause (2005) 12:117–20.[CrossRef][Web of Science][Medline]
  21. Espindola D, Kennedy KA, Fischer EG. Management of abnormal uterine bleeding and the pathology of endometrial hyperplasia. Obstet Gynecol Clin North Am (2007) 34:717–37. ix.[CrossRef][Web of Science][Medline]
  22. Ferenczy A. Pathophysiology of endometrial bleeding. Maturitas (2003) 45:1–14.[CrossRef][Web of Science][Medline]
  23. Amant F, Moerman P, Neven P, et al. Endometrial cancer. Lancet (2005) 366:491–505.[CrossRef][Web of Science][Medline]

Add to CiteULike CiteULike   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us    What's this?

Related articles in Am. J. Epidemiol.:

Risk of Complex and Atypical Endometrial Hyperplasia in Relation to Anthropometric Measures and Reproductive History
Meira Epplein, Susan D. Reed, Lynda F. Voigt, Katherine M. Newton, Victoria L. Holt, and Noel S. Weiss
Am. J. Epidemiol. 2008 168: 563-570. [Abstract] [FREE Full Text]  



This article has been cited by other articles:


Home page
 Am J EpidemiolHome page
F. Birrell and R. M. Francis
Invited Commentary: Physical Activity and Vitamin D
Am. J. Epidemiol., September 15, 2008; 168(6): 587 - 589.
[Abstract] [Full Text] [PDF]

Home page
 Am J EpidemiolHome page
M. Epplein, S. D. Reed, L. F. Voigt, K. M. Newton, V. L. Holt, and N. S. Weiss
Epplein et al. Respond to "Endometrial Hyperplasia--Getting Back to Normal"
Am. J. Epidemiol., September 15, 2008; 168(6): 575 - 576.
[Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
168/6/571    most recent
kwn165v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Related articles in Am. J. Epidemiol.
Right arrow Similar articles in this journal
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrowRequest Permissions
Right arrow Disclaimer
Google Scholar
Right arrow Articles by Lacey, J. V.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Lacey, J. V., Jr.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?