Concurrent endometrial carcinoma in women with a biopsy diagnosis of atypical endometrial hyperplasia

A Gynecologic Oncology Group study

Authors


  • See referenced commentary and companion article on pages 729–31 and 804–11, this issue.

Abstract

BACKGROUND

Adenocarcinoma of the endometrium is the most common gynecologic malignancy in the United States, accounting for approximately 36,000 diagnoses of invasive carcinoma annually. The most common histologic type, endometrioid adenocarcinoma (EC), accounts for 75–80% of patients. The objective of this work was to estimate the prevalence of concurrent carcinoma in women with a biopsy diagnosis of the precursor lesion, atypical endometrial hyperplasia (AEH).

METHODS

This prospective cohort study included women who had a community diagnosis of AEH. Diagnostic biopsy specimens were reviewed independently by three gynecologic pathologists who used International Society of Gynecologic Pathologists/World Health Organization criteria. Study participants underwent hysterectomy within 12 weeks of entry onto protocol without interval treatment. The hysterectomy slides also were reviewed by the study pathologists, and their findings were used in the subsequent analyses.

RESULTS

Between November 1998 and June 2003, 306 women were enrolled on the study. Of these, 17 women were not included in the analysis: Two patients had unreadable slides because of poor processing or insufficient tissue, 2 patients had only slides that were not endometrial, the slides for 5 patients were not available for review, and 8 of the hysterectomy specimens were excluded because they showed evidence of interval intervention, either progestin effect or ablation. In total, 289 patients were included in the current analysis. The study panel review of the AEH biopsy specimens was interpreted as follows: 74 of 289 specimens (25.6%) were diagnosed as less than AEH, 115 of 289 specimens (39.8%) were diagnosed as AEH, and 84 of 289 specimens (29.1%) were diagnosed as endometrial carcinoma. In 5.5% (16 of 289 specimens), there was no consensus on the biopsy diagnosis. The rate of concurrent endometrial carcinoma for analyzed specimens was 42.6% (123 of 289 specimens). Of these, 30.9% (38 of 123 specimens) were myoinvasive, and 10.6% (13 of 123 specimens) involved the outer 50% of the myometrium. Among the women who had hysterectomy specimens with carcinoma, 14 of 74 women (18.9%) had a study panel biopsy consensus diagnosis of less than AEH, 45 of 115 women (39.1%) had a study panel biopsy consensus diagnosis of AEH, and 54 of 84 women (64.3%) had a study panel diagnosis of carcinoma. Among women who had no consensus in their biopsy diagnosis, 10 of 16 women (62.5%) had carcinoma in their hysterectomy specimens.

CONCLUSIONS

The prevalence of endometrial carcinoma in patients who had a community hospital biopsy diagnosis of AEH was high (42.6%). When considering management strategies for women who have a biopsy diagnosis of AEH, clinicians and patients should take into account the considerable rate of concurrent carcinoma. Cancer 2006. © 2006 American Cancer Society.

Adenocarcinoma of the endometrium is the most common gynecologic malignancy in the United States, accounting for approximately 36,000 diagnoses of invasive carcinoma annually.1 The most common histologic type, endometrioid adenocarcinoma (ECa), accounts for 75–80% of diagnoses and commonly is associated with long-term, unopposed estrogenic stimulation.2 Other risk factors associated with the development of ECa include obesity, diabetes, hypertension, and diet.

It is believed that the majority of endometrioid neoplastic lesions of the endometrium follow a continuum of histologically distinguishable hyperplastic lesions that covers a spectrum ranging from endometrial hyperplasia without atypia (EH), to endometrial hyperplasia with atypia (AEH), to well differentiated ECa.3, 4 It has been proposed that a variety of histopathologic features (most notably, cytologic atypia and architectural crowding) differentiate EH from AEH and AEH from ECa. Although this continuum of endometrial hyperplasia has been accepted by the World Health Organization (WHO) and the International Society of Gynecologic Pathologists (ISGP),5 it has not undergone the same degree of rigorous, prospective, multicenter evaluation as other classification systems for preinvasive neoplastic lesions, such as cervical neoplasia.

One accepted treatment option for patients with AEH is hysterectomy with bilateral salpingo-oophorectomy. In clinical practice, however, many gynecologic oncologists have attempted conservative hormone management either to preserve fertility in younger women or to avoid hysterectomy in women who had serious comorbidities or who did not want to undergo surgery if possible. Our long-term objective was to initiate a series of trials assessing hormone, nonsurgical therapeutic regimens for the precursor lesion, AEH.

To ensure patient safety, however, we needed to ascertain the reproducibility of the diagnosis of AEH and the rate of concurrent endometrial carcinoma among women with a biopsy diagnosis of AEH: The current analysis was limited to the latter issue. Prospective data demonstrating the rate of concurrent carcinoma in women with a biopsy diagnosis of AEH are scarce. Therefore, we undertook a prospective cohort study, Gynecologic Oncology Group (GOG) 167, to estimate the prevalence of concurrent ECa in women with a community diagnosis of AEH. The study was presented to and approved by local institutional review boards (IRBs) prior to activation, and all patients provided written consent prior to enrollment on the study.

MATERIALS AND METHODS

Eligibility criteria included women with a community hospital pathology diagnosis of AEH and a GOG performance status of 0, 1, or 2. Patients who were enrolled on study underwent subsequent hysterectomy within 12 weeks of study enrollment and had no interval treatment. Patients provided written informed consent consistent with federal, state, and local requirements. The study protocol was approved by the GOG Humans Subjects Committee, which is a registered IRB, and by the individual IRBs at each participating institution.

All diagnostic biopsies subsequently were reviewed independently by three study pathologists using standard ISGP/WHO criteria (C.L.T., R.Z., and S.S.).5 Each reviewer was blinded to the diagnoses of the other two reviewers. The biopsy review was conducted before the hysterectomy review. A consensus diagnosis was defined as agreement among at least two of the three study pathologists. Hysterectomy specimens were reviewed centrally by the GOG Pathology Committee as well as by the study pathology panel, which included the pathologists who reviewed the biopsies. The review of hysterectomy specimens consisted of slides representing the most severe pathology as diagnosed by the referring institution's pathologist. The hysterectomy specimens were reviewed separately from the corresponding biopsies. The protocol mandated that either the entire hysterectomy specimen had to be submitted for review, or the slides that represented the most severe lesion, as determined by the referring institution's pathologist, had to be available for review. The findings of the study panel of pathologists formed the basis of analysis.

The biopsy review diagnoses by the study panel were subdivided into three categories: “less than” AEH, AEH, and ECa. The “less than AEH” category encompassed a spectrum of diagnoses, including hyperplasia without atypia, secretory endometrium, proliferative endometrium, benign polyps, menstrual endometrium, and inactive endometrium. In this category, the diagnosis was considered a consensus if at least two of the three reviewers diagnosed the slide as any one of those subcategories. The primary statistical outcome of this analysis was the presence of carcinoma in the hysterectomy specimen as diagnosed by the study panel.

Logistic regression analyses were performed to study the correlations between the candidate prognostic variables and each of two dichotomous outcomes: 1) the presence or absence of carcinoma in the hysterectomy specimen and 2) the presence or absence of high-risk carcinoma in the hysterectomy specimen. High-risk carcinoma was defined as myoinvasion or Grade 2–3 lesions. In both analyses, the independent variables in the logistic regression were age, biopsy diagnosis, and time from biopsy to hysterectomy Associations with other categorical factors were assessed by estimated odds ratios (ORs) and the chi-square statistic. ORs and their associated 95% confidence intervals (95% CIs) were calculated.

RESULTS

Patient Characteristics

Between November 1998 and June 2003, a total of 306 women were enrolled from 30 GOG member and affiliate member sites. The characteristics of these patients are provided in Table 1. Two patients had unreadable slides because of poor processing or insufficient tissue, two patients had specimens that were not from the endometrium, and five patients did not have material available for review. Eight hysterectomy specimens were excluded, because they showed evidence of interval intervention, either progestin effect or ablation. In total, 289 patients were included in this analysis.

Table 1. Characteristics of Women with a Consensus Diagnosis of Atypical Endometrial Hyperplasia
CharacteristicCarcinoma in hysterectomy specimen, no. of patients
NoYes
  1. D & C: dilatation and curettage.

Age  
 ≤ 40 yrs88
 41–50 yrs4726
 51–60 yrs5534
 61–70 yrs3532
 71–80 yrs1517
 > 80 yrs66
 Median (yrs)5559
Ethnicity
 African American99
 Caucasian139105
 Hispanic112
 Asian45
 Unknown30
 Other02
Method of payment
 Private insurance9960
 Medicare2518
 Medicaid127
 Medicare/Medicaid42
 Medicare/private714
 Self pay38
 No means of pay71
 Other government810
 Veterans administration11
 Unknown02
How biopsy obtained
 D & C8028
 Novak1616
 Vabra6878
 Pipelle21

The mean age of the study participants was 58 years (range, 25–89 yrs). The ethnic distribution included 18 African-American women (6.2%), 244 non-Hispanic white women (84.4%), 13 Hispanic women (4.5%), 9 Asian women (3.1%), 3 women of unknown ethnicity (1.0%), and 2 women of “other” ethnicity (0.7%). The median time from biopsy to hysterectomy was 6 weeks (range, 0–53 weeks).

The study panel consensus diagnosis for the biopsies was “less than AEH” in 74 of 289 specimens (25.6%), AEH in 115 of 289 specimens (39.8%), carcinoma in 84 of 289 specimens (29.1%), and there was no consensus in 16 of 289 specimens (5.5%) (Fig. 1).

Figure 1.

Results from the study panel review of diagnostic hysterectomy biopsy specimens is illustrated. AEH: atypical endometrial hyperplasia.

The hysterectomy review showed endometrial carcinoma in 42.6% of specimens (Fig. 2). The number of patients who had ECa in their hysterectomy specimens was 14 of 74 patients (18.9%) who had a study panel consensus diagnosis of “less than AEH”, 45 of 115 patients (39.1%) who had a study panel consensus diagnosis of AEH, 54 of 84 patients (64.3%) who had a study panel consensus diagnosis of carcinoma, and 10 of 16 patients (62.5%) who had no study panel consensus (Table 2). The carcinomas that were diagnosed in hysterectomy specimens ranged in severity from Stage IA, Grade 1 lesions; to Stage IB, Grade 3 lesions; to Stage IC, Grade 1 lesions. Although the great majority of the tumors were Grade 1, 6 tumors were Grade 2, and 2 tumors were Grade 3. No patients with Stage II, III, or IV carcinoma were identified (Table 4). Of the carcinomas identified, 80 of 123 tumors (65.0%) were confined to the endometrium. However, 38 of 123 tumors (30.9%) were myoinvasive, and 13 of 123 tumors (10.6%) involved the outer 50% of the myometrium.

Figure 2.

The prevalence of concurrent carcinoma in hysterectomy specimens (SP) is illustrated. AEH: atypical endometrial hyperplasia.

Table 2. Histology Review and Associated Effects
VariableHysterectomy reviewOR relative to AEH (95% CI)P
No carcinomaCarcinoma
  1. OR: odds ratio; 95% CI: 95% confidence interval; AEH: atypical endometrial hyperplasia.

Carcinoma in hysterectomy specimen    
 Biopsy    
  < AEH60140.380 (0.189–0.767)0.0069
 Review    
  AEH7045  
  Carcinoma30542.864 (1.586–5.171)0.0005
  No consensus6102.528 (0.852–7.506)0.0948
High-risk carcinoma in hysterectomy specimen    
 Biopsy    
  < AEH7310.098 (0.013–0.761)0.0263
 Review    
  AEH9817  
  Carcinoma66181.625 (0.762–3.468)0.2092
  No consensus974.400 (1.390–13.934)0.0118
Table 3. Hysterectomy Specimens with Carcinoma
VariableOR (95% CI)P
  1. OR: odds ratio; 95% CI: 95% confidence interval.

Carcinoma in hysterectomy specimen  
 Weeks from biopsy to hysterectomy0.949 (0.903–0.997)0.0374
 Age0.949 (0.903–0.997)0.4121
High-risk cancer in hysterectomy specimen  
 Weeks from biopsy to hysterectomy0.916 (0.838–1.002)0.0560
 Age1.047 (1.015–1.079)0.0038

High-risk features, which were defined as myoinvasion or Grade 2 or 3 lesions, were present in 43 tumors. Among the patients who had carcinoma with high-risk features diagnosed in their hysterectomy specimen, 1 patient (2.3%) with a Grade 1, Stage IB lesion had a biopsy diagnosis of “less than AEH”; 17 patients (39.5%) were diagnosed with AEH at biopsy; 18 patients (41.9%) had a biopsy diagnosis of carcinoma; and 7 patients (16.3%) had no consensus at biopsy. Conversely, in the 78 patients who had biopsy results that were interpreted as “less than AEH” by the study panel, only 1 patient had a carcinoma with high-risk features found in her hysterectomy specimen: a Grade 1, Stage IB lesion.

For the 14 patients who had a biopsy diagnosis of “less than AEH” and carcinoma in their hysterectomy specimen, 1 patient had a diagnosis of “proliferative endometrium,” 2 patients were “menstrual,” 2 patients were diagnosed with “normal endometrium,” and 2 patients were diagnosed as “other normal endometrium.” Two patients were diagnosed with polyps, two patients were diagnosed with “simple EH,” and three patients were diagnosed with “complex EH.” In two patients, the study pathologists did not reach a consensus other than “less than AEH.”

The degree to which the study panel's consensus biopsy diagnoses were predictive of carcinoma in the hysterectomy specimen is presented in Table 2. The OR for patients who had a biopsy diagnosis upgraded by the study panel to carcinoma, relative to patients who were diagnosed with AEH, was 2.864 (95% CI, 1.586–5.171; P = 0.0005). The OR for patients who had a biopsy diagnosis downgraded to “less than AEH” by the study panel relative to patients who were diagnosed with AEH was 0.380 (95% CI, 0.189–0.767; P = 0.0069). Among patients for whom the panel did not reach a consensus on the biopsy specimen, the OR was 2.528 (95% CI, 0.852–7.506; P = 0.0948). Although the effect size was not large, elapsed time from biopsy to hysterectomy reached conventional levels of statistical significance (OR, 0.949; 95% CI, 0.903–0.997; P = 0.0374), but age was not a significant factor (OR, 1.009; 95% CI, 0.988–1.030; P = 0.4121) (Table 3).

Table 4. Endometrioid Endometrial Carcinoma Diagnosed in Hysterectomy Specimens
VariableNo. of specimens
Grade 1 (n = 115)Grade 2 (n = 6)Grade 3 (n = 2)
No stageStage IAStage IBStage ICNo stageStage IAStage IBStage ICNo stageStage IAStage IBStage IC
  1. AEH: atypical endometrial hyperplasia.

< AEH0131000000000
AEH2266711000110
Carcinoma13511511000000
No agreement034100200000
Total377221322200110

The distribution according to age and ethnicity of the patients who had carcinoma in their hysterectomy specimen did not differ significantly from the characteristics of our cohort as a whole or from patients in the study who did not have carcinoma in their hysterectomy specimen. The median age of patients who had hysterectomy specimens with carcinoma was 59 years (range, 25–83 yrs), and the median age of patients who did not have carcinoma was 55 years (range, 29–89 yrs). The median time to resection for both groups was similar: 6.3 weeks for patients who were diagnosed with carcinoma and 7.9 weeks for patients who had no carcinoma identified. The ethnic distribution and the methods of payment among the two subgroups were similar.

High-Risk Carcinomas

The study panel consensus diagnoses for biopsies also were predictive of high-risk carcinoma in hysterectomy specimens. The OR for patients who had a biopsy diagnosis upgraded to carcinoma by the study panel, relative to patients who were diagnosed with AEH, was 1.625 (95% CI, 0.762–3.468; P = 0.2092). The OR for patients who had a biopsy diagnosis downgraded to “less than AEH” by the study panel, relative to patients who had AEH, was 0.098 (95% CI, 0.013–0.761; P = 0.0263). Patients who had no biopsy consensus diagnosis had an OR of 4.4 (95% CI, 1.390–13.934; P = 0.0188). The elapsed time from biopsy to hysterectomy was not statistically significant (OR, 0.0916; 95% CI, 0.838–1.002; P = 0.0560), but age was a significant factor (OR, 1.047; 95% CI, 1.015–1.079; P = 0.0038).

DISCUSSION

This prospective cohort study was designed to estimate the prevalence of concurrent carcinoma in patients who have a biopsy diagnosis of atypical endometrial hyperplasia, which is believed to be the immediate precursor lesion to endometrioid endometrial carcinoma.4 We found that the prevalence of carcinoma in hysterectomy specimens was 42.6%. Carcinoma was significantly more likely if the study panel review consensus diagnosis of the biopsy was upgraded to carcinoma (OR, 2.864; 95% CI, 1.568–5.171; P = 0.0005) than if the review was interpreted as “less than AEH” (OR, 0.380; 95% CI, 0.189–0.767; P = 0.0069). However, even the subset of women who had biopsies that were interpreted as “less than AEH” by the GOG study panel had an 18.9% prevalence of concurrent carcinoma. Conversely, among patients who had an upgraded biopsy diagnosis of carcinoma, no carcinoma was identified in the hysterectomy specimen in 30 of 84 patients (35.71%).

Cullen posited an etiologic correlation between endometrial hyperplasia and endometrial carcinoma more than 100 years ago.6 Subsequent reviews by Taylor in 1932 and by Novak and Yui in 1936 supported this observation.7, 8 In 1947, Gusberg focused attention on the role of estrogenic stimulation as a factor that caused endometrial hyperplasia and carcinoma.9 A variety of classification systems for endometrial hyperplasia have been proposed over the last 50 years.10–17 A group of gynecologic pathologists at Stanford University, for example, proposed a classification algorithm for predicting clinically significant endometrial lesions.18 That algorithm was predicated on the use of 1 architectural measure combined with 2 cytologic parameters, and it had 99.5% specificity and 57% sensitivity for detecting myoinvasion based on a set of clinical specimens that were obtained at community hospitals. The WHO endorsed a classification system for endometrial hyperplasia in 1975 that was revised in 1994 and again in 2003. Several expert panels have noted considerable interobserver and intraobserver variability with the use of these systems, underscoring the difficulty of making the diagnosis of premalignant changes of the endometrium on biopsy specimens.19–21

The overall rate of concurrent carcinoma in our study was 42.6%. The rate of concurrent carcinoma in patients who had their biopsy diagnosis upgraded to carcinoma by the study panel was 64.3%. This rate was significantly higher than the rate of concurrent carcinoma in patients who had their biopsy diagnosis downgraded to “less than AEH” (18.9%) by the study panel. The study panel biopsy reviews were conducted independently and before the hysterectomy reviews. The fact that the study panel consensus diagnosis downgraded severity nearly as often as it upgraded severity suggests a lack of bias. Although each patient in this study had been diagnosed with AEH by the referring institution's pathologist, the consensus diagnosis by the study panel based on the biopsy specimen was more predictive of the hysterectomy diagnosis.

In the current study, the pathologists who made the diagnosis of AEH in the community were familiar with GOG classifications for gynecologic malignancies, which follow WHO and ISGP guidelines. For eligibility evaluation, as part of routine GOG quality assurance, all histologic specimens are required to undergo central pathology review. The patient volume in 2002 for all GOG clinical trials was 2407 patients, with a rejection rate between 3% and 4%. It is noteworthy that, in this study, the panel of 3 pathologists concurred with the referring diagnosis in only 39% of patients. A discrepancy rate between GOG pathologists and referring pathologists of 61% is very unusual. An analysis of reasons for discrepancies will be presented separately.

Several groups have reported that the prevalence of concurrent carcinoma among patients with AEH on biopsy ranged between 17% and 52%12, 15, 16, 22, 23 (Table 5). In our study, we observed a prevalence of 42.6% in patients who had a community diagnosis of AEH. Several factors may have contributed to the fact that the prevalence of concurrent carcinoma in our study cohort was relatively high compared with that reported in the literature. First, in contrast to previously published work, the current study was a prospective, multiinstitutional effort with independent pathology review by a panel of gynecologic pathologists. Second, the study sample was an order of magnitude larger than the other studies. This study also differed from others in that we found no significant association of carcinoma with patient age. This surprising finding may be have been because of a selection bias, in that younger women who were interested in participating in research to assess nonsurgical therapies may have been represented disproportionately in our cohort. It may be noteworthy that the prevalence of carcinoma reported here, although it was much higher than that reported in previously studies, was similar to that reported by the two most recent series. The threshold for the diagnosis of carcinoma may have been lowered in recent years, particularly in hysterectomy specimens. Conversely, the threshold for the diagnosis of carcinoma may be higher in biopsy specimens, because such a diagnosis ordinarily would prompt major surgery. It also should be noted that discussions of progression of atypical hyperplasia to carcinoma in the older literature should be interpreted with caution, because these clinical situations actually may represent persistence of carcinomas that already were extant at the time of the initial diagnosis of atypical hyperplasia.

Table 5. Atypical Endometrial Hyperplasia and Concurrent Endometrioid Carcinoma of the Uterus
StudyType of studyNo. of patientsPercentage of patients with concurrent carcinomaPercentage of patients with myoinvasion
Gusberg and Kaplan, 196312Retrospective, single-institution case series1820.0Not recorded
Tavassoli and Kraus, 197815Retrospective, single-institution case series4825.02.1
Kurman and Norris, 198216Retrospective, referral case series8917.07.9
Janicek and Rosenshein, 199422Retrospective, single-institution case series4443.039.0
Dunton et al., 199623Retrospective, single-institution case series2352.026.0
Current studyProspective, multiinstitutional cohort28943.013.2

The stage and grade of the carcinomas identified in this cohort must be taken into account when planning treatment strategies. Although 77 of 123 carcinomas (63%) that we identified were Grade 1 lesions confined to the endometrium, 31% were myoinvasive, and 11% involved the outer 50% of the myometrium. Even among the patients who had their biopsy diagnosis downgraded by the study panel, the prevalence of concurrent carcinoma was 18.9%. The study panel was not infallible: in a full 36% of patients for whom the panel upgraded the diagnosis in the biopsy specimen, no carcinoma was identified in the hysterectomy specimen. Although these discrepancies raise questions of sampling error, in several patients, the study panel did predict that, in the hysterectomy specimen, there would be no carcinoma found, because lesions that had been diagnosed as probable carcinoma were very focal in a background of AEH, or they appeared to be confined to a polyp.

Developments in our understanding of the molecular events in preinvasive endometrial disease, combined with quantitative imaging analysis algorithms, have permitted a more quantitative, less subjective analysis of various hyperplastic endometrial lesions. For example, the lesions that carry a greater risk of disease progression to carcinoma appear to be monoclonal, with early loss of functional PTEN protein expression.24 We hypothesize that true malignant changes initially appear focally and subsequently expand. These changes are in contrast to those seen as sequelae of benign phenomena, which are manifested as more of a field effect. The Endometrial Collaborative Group has proposed a new term, “endometrial intraepithelial neoplasia” (EIN),4 to characterize early malignant lesions. Even EIN, however, has not yet undergone rigorous prospective evaluation or determinations of reproducibility.

Future studies that use the material from this study will include histomorphometric imaging and biomarker analysis in an effort to characterize endometrial lesions in biopsy specimens that are predictive of carcinoma at the time of immediate hysterectomy.25, 26 We also plan to explore potential explanations for diagnostic discrepancies. In the current study, many of the biopsy specimens contained scant material, and some were processed poorly. It is possible that we may need to determine what constitutes an “adequate” specimen, which has been done with cervical Papanicolaou smears. Finally, in addition to issues of reproducibility, the clinical significance of the current paradigm of endometrial hyperplastic, or preinvasive, lesions should be rethought. The prognosis for patients with Stage IA, Grade 1 ECa is good. We may want to consider whether all well differentiated endometrioid ECa and AEH should be diagnosed as endometrial neoplasia, as proposed by Bergeron et al.20 This distinction between what we think of as AEH and well differentiated ECa may be one that cannot be made on endometrial samples; in fact, it may not be important biologically. Currently, however, prospective trials of conservative hormone regimens will need to take into account the baseline rate (42.6%) of ECa among women who have a diagnosis of AEH both in trial design and in appropriate patient education.

Acknowledgements

The following Gynecologic Oncology Group member institutions participated in this study: Duke University Medical Center, Walter Reed Army Medical Center, University of Mississippi Medical Center, University of Pennsylvania Cancer Center, Milton S. Hershey Medical Center, University of North Carolina School of Medicine, University of Iowa Hospitals and Clinics, University of Texas Southwestern Medical Center at Dallas, Wake Forest University School of Medicine, University of California Medical Center at Irvine, Tufts-New England Medical Center, University of Kentucky, The Cleveland Clinic Foundation, State University of New York at Stony Brook, Southwest Oncology Group, Washington University School of Medicine, Memorial Sloan-Kettering Cancer Center, Columbus Cancer Council, University of Massachusetts Medical School, Fox Chase Cancer Center, Women's Cancer Center, University of Oklahoma, University of Virginia, University of Chicago, Tacoma General Hospital, Brookview Research Inc., Ellis Fischel Cancer Center, Fletcher Allen.

Ancillary