Article Text


Web based pathology assessment in RTOG 98-04
  1. Wendy A Woodward1,
  2. Nour Sneige1,
  3. Kathryn Winter2,
  4. Henry Mark Kuerer1,
  5. Clifford Hudis3,
  6. Eileen Rakovitch4,
  7. Barbara L Smith5,
  8. Lori J Pierce6,
  9. Isabelle Germano7,
  10. Anthony T Pu8,
  11. Eleanor M Walker9,
  12. David Lawrence Grisell10,
  13. Julia R White11,
  14. Beryl McCormick3
  15. for the Radiation Therapy Oncology Group (RTOG)
  1. 1University of Texas-MD Anderson Cancer Center, Houston, Texas, USA
  2. 2RTOG Statistical Center, Philadelphia, Pennsylvania, USA
  3. 3Memorial Sloan-Kettering Cancer Center, New York, New York, USA
  4. 4Sunnybrook Health Sciences Centre, Toronto, Ontario, Canada
  5. 5Massachusetts General Hospital, Boston, Massachusetts, USA
  6. 6University of Michigan Comprehensive Cancer Center, Ann Arbor, Michigan, USA
  7. 7Mount Sinai Medical Center, New York, New York, USA
  8. 8Radiological Associates of Sacramento, Sacramento, California, USA
  9. 9Henry Ford Hospital, Detroit, Michigan, USA
  10. 10Cancer Centers of the Carolinas-Greenville CCOP, Greenville, South Carolina, USA
  11. 11Stephanie Spielman Comprehensive Breast Center, Columbus, Ohio, USA
  1. Correspondence to Dr Beryl McCormick, Memorial Sloan-Kettering Cancer Center, 1275 York Avenue, New York, NY 10065, USA; mccormib{at}


Aims Radiation Therapy Oncology Group 98-04 sought to identify women with ‘good risk’ ductal carcinoma in situ (DCIS) who receive no significant benefit from radiation. Enrolment criteria excluded close or positive margins and grade 3 disease. To ensure reproducibility in identifying good risk pathology, an optional web based teaching tool was developed and a random sampling of 10% of submitted slides were reviewed by a central pathologist.

Methods Submitting pathologists were asked to use the web based teaching tool and submit an assessment of the tool along with the pathology specimen form and DCIS H&E stained slide. Per protocol pathology was centrally reviewed for 10% of the cases.

Results Of the 55 DCIS cases reviewed, three had close or positive margins and three were assessed to include grade 3 DCIS, therefore 95% of DCIS cases reviewed were correctly graded, and 89% reviewed were pathologically appropriate for enrolment. Regarding the teaching tool, 13% of DCIS cases included forms that indicated the website was used. One of these seven who used the website submitted DCIS of grade 3.

Conclusions Central review demonstrates high pathological concordance with enrolment eligibility, particularly with regard to accurate grading. The teaching tool appeared to be underused.

  • Breast
  • Breast Cancer
  • Breast Pathology

This is an Open Access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 3.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See:

Statistics from


With the growing acceptance of mammography as a screening tool, non-palpable ductal carcinoma in situ (DCIS) is being diagnosed with a frequency of 20–25% in large breast practices. At the time Radiation Therapy Oncology Group (RTOG) 98-04 opened, retrospective studies suggested that good risk disease, that is, small lesions, with a low-grade pathology classification, might be effectively treated with radiation or with observation.1–3 Among randomised trials studying the benefit of postlumpectomy radiation therapy for DCIS, all ultimately demonstrated a benefit to postlumpectomy radiation for all cohorts.4–15 Around the time the RTOG 98-04 was designed, two of these had been reported.9 ,11 The National Surgical Adjuvant Breast and Bowel Project (NSABP) B-17 trial stratified patients by age, presence or absence of lobular carcinoma in situ, presence or absence of an axillary node dissection, and method of detection, that is, palpable mass versus mammogram only. Size, margins and pathological grade, recognised ultimately as important selection factors for possible treatment with observation only, were not factors in that trial.9 The NSABP B-24 trial, placed all patients on radiation and then randomised them to postradiation observation or tamoxifen.11 Thus, neither trial addressed the question of the efficacy of no radiation in a subset of good-risk patients with DCIS. RTOG 98-04 was designed to address this gap in knowledge.

Multiple classification systems for DCIS have been proposed to assess prognostic factors such as size, extent of disease, margins and pathology grade; acceptance of uniform criteria for these factors, however, has not been achieved. Prior to the initiation of RTOG 98-04, a consensus conference was held with the purpose of defining pathological criteria for this disease. A consensus was reached by the participating pathologists and this working system was a starting point to assure standardisation in defining cases of pathological ‘good-risk’ disease. Using this consensus,16 a ‘teaching’ set of photomicrographs illustrating key diagnostic features of low, intermediate and high grade lesions was prepared and made available on the RTOG website as a resource for the designated pathologist from participating institutions. The consensus16 forms the basis of the current DCIS grading guidelines endorsed by the College of American Pathologists (CAPs), and this RTOG teaching website is referenced in the CAP guidelines.17 In RTOG 98-04, mandatory credentialing was not a requirement for study participation. However, slides from a random sample of 10% of patients on the study were retrospectively collected and reviewed by a dedicated breast pathologist. Herein the results of this sample review and the use of the teaching tool are reported.


The definition of ‘good risk’ DCIS used the guidelines accepted at a classification consensus conference of DCIS.16 Patients were ineligible if the DCIS was high grade, >2.5 cm in greatest diameter as measured on the preoperative mammogram or pathological specimen, or had final pathological margin <3 mm. High histological grade was defined by the presence of either of two features: (1) high nuclear grade, defined as nuclear size >2.5 times lymphocyte or benign ductal epithelial cell nuclei, marked pleomorphism, irregular chromatin with single or multiple nucleoli and conspicuous mitoses, or (2) necrosis, defined by the presence of ghost cells with karyorrhectic debris, in a third or greater of the neoplastic ducts.

Specimen handling recommendations provided in RTOG 98-04 are described in online supplementary methods (document 1). Grading criteria were explicitly defined. Low grade nuclei (NG 1) applied to lesions with monotonous nuclei, size 1.5–2.0 normal red blood cell (RBC) or duct epithelial cell nucleus dimensions, which usually exhibit diffuse, finely dispersed chromatin with only occasional nucleoli and mitotic figures. NG 1 was further defined as usually associated with polarisation of constituent cells. Importantly, the presence of nuclei that were of similar size but were pleomorphic precluded a low grade classification. High-grade nuclei (NG 3) classification was applied to nuclei that were markedly pleomorphic, usually >2.5 RBC or duct epithelial cell nuclear dimensions, and usually demonstrated vesicular and irregular chromatin distribution and prominent, often with multiple nucleoli mitosis may be conspicuous. Nuclei that were neither NG 1 nor NG 3 were to be assigned to intermediate grade nuclei (NG 2). Any central zone necrosis within a duct, usually exhibiting a linear pattern within ducts if sectioned longitudinally was considered comedonecrosis. Punctate necrosis was defined as non-zonal type necrosis, defined as foci of individual cell necrosis visible under 10×. The required microscopic examinations included: (1) extent of DCIS (determined by number of sections containing DCIS as well as the largest dimension of DCIS lesion on a glass slide) (2) nuclear grade, (3) necrosis, (4) margins of resection (closest ≥3–9 mm, ≥10 mm or a re-excision margin), (5) cell polarisation, (6) architectural patterns, and (7) the relationship of calcifications, when present, to the DCIS.

In order to improve reproducibility of histological criteria, a ‘teaching’ set of photomicrographs illustrating key diagnostic features of low/intermediate and high grade lesions was prepared by the designated RTOG breast central pathologist (NS) and made available on the RTOG website ( Of note, this teaching set was selected from a group of DCIS cases that were the subject of a consensus review and all grading had been agreed upon by all participants from the study.15 This was provided as a resource for the designated pathologist from participating institutions; however, mandatory credentialing was not a requirement for study participation. A random sampling of 10% of cases (n=62) were requested for review by the central RTOG breast pathologist for the study (NS) to assess the success of the teaching tool in educating pathologists in the assignment of low or intermediate histological grade as defined in this protocol. Standard pathology submission forms as well as a teaching tool assessment form were expected to be submitted with the slides and were reviewed in the central sampling review. Pathology discrepancies identified did not render patients ineligible.


Results are summarised in figure 1. Of the 62 requested cases, correct slides were submitted for 60, of which 5/60 (8%) had no DCIS on the slide. In three cases, slides submitted were from negative re-excisions and documented adequate margin per protocol. Fifty-two of 55 cases (95%) with DCIS slides available for review were grade 1 or grade 2 DCIS. Three of 55 (5%) were grade 3 DCIS. Of the 52 grade 1/2 DCIS cases, 3 had close or positive margins, therefore 49/55 cases (89%) reviewed were pathologically appropriate for enrolment based on the data available for review.

Figure 1

Summary of central review findings GI, gastrointestinal; DCIS, ductal carcinoma in situ.

Regarding the teaching tool, 7/55 (13%) cases with DCIS slides included the forms intended to be returned with the slides that included responses that indicated the website was used. Six of these seven submitted DCIS of grade 1 or 2. One of the three slides that were found to contain grade 3 DCIS on central review included documentation that the website was used.


Grading criteria for DCIS are complex and have changed over time. Uniform, reproducible grading criteria are critical for clinical trials basing randomisation and treatment strategy on these findings. State of the science guidelines were provided to participants in the protocol for RTOG 98-04 and on central review, 95% of cases were correctly graded as grade 1 or grade 2. While significant efforts were made to provide additional tools to pathologists to ensure accurate uniform grading, these appear to be underused and thus a direct benefit for this intervention could not be measured. Strategies for increasing awareness and use of similar tools in future studies are needed.

While nuclear grade and comedonecrosis have been strongly associated with increased risk of ipsilateral breast recurrence after lumpectomy alone in many studies,4 ,14 ,18 heterogeneity of populations and methods make consensus regarding true risk elusive.2 ,19 Length of follow-up may be a critical factor in this issue as high-grade lesions have been reported to recur earlier20 Four previously reported randomised trials comparing lumpectomy alone with lumpectomy and whole breast radiation have recently been reviewed by the Early Breast Cancer Trialists’ Collaborative Group.7 This overview included trials by the NSABP9 ,10 ,18 and the European Organization for Research and Treatment of Cancer4–6 ,13 which centrally reviewed >75% of cases, as well as the SweDCIS (Swedish DCIS)8 ,14 and UK/Australia New Zealand DCIS,12 which centrally reviewed 25% and 0%, respectively. While not compared directly, they report that among 1794 patients treated with lumpectomy alone, the 10-year ipsilateral breast cancer recurrence was 21.7%, 24.8% and 32.2% for DCIS histological grade low, intermediate and high, respectively.7 Among 1617 women treated with lumpectomy alone evaluated by nuclear grade, the 10-year ipsilateral breast cancer recurrence was 28.4%, 29.8% and 33.1% for DCIS nuclear grade low, intermediate and high, respectively. As reported in each trial individually, in all cohorts there was a significant benefit to radiotherapy.7 In contrast, recent retrospective series from large centres failed to show an association between grade and presence of necrosis with risk of local failure.21–23 However these cohorts included patients who received radiotherapy and the Early Breast Cancer Trialists’ Collaborative Group meta-analysis suggests equal benefit among patients of all grades who receive radiation potentially confounding these results. While many factors may contribute to these heterogeneous results, standardisation of pathological assessment and consistent use of grading systems across samples is critical to understanding these contributions moving forward.

Although it has not been proven, it is possible that misclassification of grading contributes to variations in the reported risk associated with grade. Few studies have rigorously examined concordance between pathologists specifically in applying grading criteria for DCIS. Ringberg et al14 report that the agreement between pathologists was moderate (κ=0.486) in the SweDCIS trial. Correctness of diagnosis in the subcohort of SweDCIS was 84.8%. In another such study by Sneige et al,15 six surgical pathologists from four institutions used the Lagios grading system to grade 125 DCIS lesions. Before meeting to evaluate the cases, a training set of 12 glass slides, including cases chosen to present conflicting cues for classification, was mailed to the participants with a written criteria summary. This was followed by a working session in which criteria were reviewed and agreed on. The pathologists then graded the lesions independently. A complete agreement among raters was achieved in 43 (35%) cases, with five of six raters agreeing in another 45 (36%) cases. In no case did two raters differ by more than one grade. Generalised κ value similarly indicated moderate agreement (0.46, SE=0.02). The authors conclude that with adherence to specific criteria, interobserver reproducibility in the classification of DCIS cases can be obtained in most cases.

By comparison, the pathological assignment of grade in RTOG 98-04 was accurate in 95% of cases, representing excellent concordance. While the pathology tools appeared underused, it is difficult to ascertain whether multiple cases submitted by the same pathologist or institution may have been documented as using the website for the first case and not for subsequent cases. The tool remains an inexpensive, scalable, easily accessible tool online, and may be of use in future studies of intraobserver grading after utilisation. The 5% inaccuracy remains important for this common disease when treatment decisions are based on it. In addition, there is no way to estimate the number of patients inaccurately deemed ineligible based on misclassifying low/intermediate grade lesions as high grade. Finally, of course, grade is not the only criteria needed to accurately assign good risk DCIS. Incorporating data reviewed regarding margin status, the total correctness of diagnosis in the cohort of RTOG 98-04 is 89%, although this may be a low estimate given that the submitted slides may not represent the final margins achieved.

Overall, pathological accuracy on RTOG 98-04 was excellent, but room for improvement remains. Recently the CAP has published guidelines that will assist pathologists in providing clinically useful and relevant information when reporting results of surgical specimen examinations.17 Moving forward, consideration for requiring use of available tools or real-time central pathology review may be warranted to achieve complete compliance. Clearly, DCIS is a heterogeneous disease, and the ability to predict ‘good risk’ DCIS depends on reproducible clinicopathological and potentially molecular assessments to best select therapy for these patients.

Take home messages

  • Overall, pathological accuracy on RTOG 98-04 was excellent.

  • Although underutilized here, the tool developed for RTOG 98-04 remains an inexpensive, scalable, easily accessible Online teaching aid, and may be of use in future studies of intraobserver grading.


View Abstract
  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:


  • Contributors The authorship has been reviewed by the RTOG and conforms to the policies of the cooperative group that ran the trial. Specifically: HMK, CH, ER, BLS, LJP, IG, ATP, EMW and DLG contributed to the design and success of the clinical trial from which this correlative report derives and carefully reviewed and contributed to the content of the manuscript. WAW drafted the manuscript and analysed data. NS performed all pathological assessments and reviewed the manuscript. KW performed statistical review, identified cases and reviewed the manuscript. JRW and BMC were primary drafters of the protocol, data collection, oversaw the entire project and contributed to drafting and reviewing the manuscript.

  • Funding This project was supported by RTOG grants U10 CA21661 and CCOP grant U10 CA37422 from the National Cancer Institute (NCI).

  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.