Abstract

Abstract 4263

Background:

COMPLETE is a registry of peripheral T-cell lymphoma (PTCL) patients from academic and community settings in the US. For the registry, selected biomarkers are recorded to determine which markers may have informed the diagnosis of PTCL. Research staff at participating sites enter data from pathology reports into an electronic data capture (EDC) system; however, pathology reports can be difficult to interpret, particularly for those who have not received specialized training. To ensure that biomarker data in COMPLETE accurately reflects pathology reports, a biomarker QA initiative was undertaken. A similar initiative was pursued by the International T-Cell Project and is the subject of a separate, parallel abstract.

Methods:

The COMPLETE biomarker QA initiative includes review of on-site data entered for 26 biomarkers commonly assessed in the diagnostic work-up of PTCL. External review by a board-certified pathologist is triggered once a site has entered the biomarker data into the EDC system for enrolled patients and uploaded the de-identified pathology reports. All patients enrolled at a site within 30 days of the first patient were reviewed. A single mismatch between the site-entered data and the reviewer's findings counts as an error. Findings of the review are entered into the EDC system to facilitate analysis and queries are sent to the site in cases where corrections are needed.

Results:

Data for 119 patients from 43 centers comprising a sum of 3570 biomarker entries were reviewed by an external hematopathologist. The mean number of phenotypic markers assessed per patient was 12 (range: 4–20), and the mean number of gene rearrangement tests performed was 0.2 (range: 0–3). Of the 119 patients, 13 (11%) had no EDC entry errors, 33 (28%) had 1–2 errors, 34 (29%) had 3–5 errors, 31 (26%) had 6–10 errors and the remaining 8 cases (7%) had 11–15 errors. The reviewer agreed with 74% of entries where the site noted a finding was positive; of the remaining entries where the site indicated the findings were positive, the reviewer noted they were negative (12%), not assessed (10%) or indeterminate (5%). For entries where the site indicated the finding was negative, the reviewer was in agreement 79% of the time; for the remaining 21% of negative-reported entries by the site, the reviewer indicated the marker was not assessed (12%), indeterminate (6%) or positive (3%). The reviewer was in agreement with the site in 92% of the entries where the site indicated the marker was not assessed; otherwise, the reviewer found the marker was negative (4%), positive (3%) or indeterminate (1%). The markers accounting for the most errors (each marker accounting for 5% or more of errors) were CD8, CD20, CD30, Epstein-Bar Virus (EBV) and T-cell receptor gamma gene rearrangement (TCR-g). The types of errors for these markers are noted in the table below.

Marker, N (%) Errors OverallType of Error Noted by the Reviewer for Markers with the Most Errors, n (%)
Finding is recorded in non-neoplastic cellsPathology report unclearTest performed, result not in pathology reportMarker is positive or negative; site notes oppositeMarker is not assessed or indeterminate; site notes positive or negative resultMarker is positive or negative; site notes marker was not assessedOther
CD8, 40 (8%) 1 (3%) 12 (30%) 2 (5%) 11 (28%) 7 (18%) 6 (15%) 1 (3%) 
CD20, 47 (9%) 24 (51%) 6 (13%) 0 (0%) 5 (11%) 6 (13%) 6 (13%) 0 (0%) 
CD30, 25 (5%) 5 (20%) 3 (12%) 1 (4%) 3 (12%) 6 (24%) 7 (28%) 0 (0%) 
EBV, 27 (5%) 1 (4%) 2 (7%) 0 (0%) 0 (0%) 0 (0%) 23 (85%) 1 (4%) 
TCR-g, 24 (5%) 0 (0%) 4 (17%) 0 (0%) 0 (0%) 5 (21%) 12 (50%) 3 (13%) 
Marker, N (%) Errors OverallType of Error Noted by the Reviewer for Markers with the Most Errors, n (%)
Finding is recorded in non-neoplastic cellsPathology report unclearTest performed, result not in pathology reportMarker is positive or negative; site notes oppositeMarker is not assessed or indeterminate; site notes positive or negative resultMarker is positive or negative; site notes marker was not assessedOther
CD8, 40 (8%) 1 (3%) 12 (30%) 2 (5%) 11 (28%) 7 (18%) 6 (15%) 1 (3%) 
CD20, 47 (9%) 24 (51%) 6 (13%) 0 (0%) 5 (11%) 6 (13%) 6 (13%) 0 (0%) 
CD30, 25 (5%) 5 (20%) 3 (12%) 1 (4%) 3 (12%) 6 (24%) 7 (28%) 0 (0%) 
EBV, 27 (5%) 1 (4%) 2 (7%) 0 (0%) 0 (0%) 0 (0%) 23 (85%) 1 (4%) 
TCR-g, 24 (5%) 0 (0%) 4 (17%) 0 (0%) 0 (0%) 5 (21%) 12 (50%) 3 (13%) 

Sum may exceed 100% due to rounding.

Recording the type of EBV test performed [in situ hybridization (ISH) and immunohistochemistry (IHC)] presented further challenges, accounting for 8% of errors. Common errors included: 1) the site indicated the test was not performed when it was performed per report and 2) the site indicated a particular test (ISH, IHC) was performed but the method used was unclear in the report.

Conclusions:

Results of the COMPLETE biomarker QA analysis indicate that errors in interpretation or recording of biomarker data from pathology reports are common. For registries and other observational studies, periodic reviews by an external pathologist may guide site training and improve the accuracy of data collection. For interventional clinical trials in PTCL, central review of all biomarker data by expert hematopathologists should be considered.

Disclosures:

Hsi:Allos: Research Funding; Eli Lilly: Research Funding; Abbott: Research Funding; Cellerant Therapeutics: Research Funding; BD Biosciences: Research Funding; Millenium: Research Funding. Gruver:MedNet Solutions: Honoraria. Foss:Allos: Consultancy; Celgene: Consultancy; Seattle Genetics: Consultancy; Eisai: Consultancy; Celgene: Study Grant, Study Grant Other; Merck: Study Grant, Study Grant Other. Carson:Allos: Consultancy, Speakers Bureau; Celgene: Consultancy, Speakers Bureau. Pinter-Brown:Allos: Consultancy, Membership on an entity's Board of Directors or advisory committees. Horwitz:Allos: Consultancy, Research Funding. Rosen:Allos: Consultancy, Honoraria. Pro:Celgene: Honoraria, Research Funding; Spectrum : Honoraria; Allos: Honoraria; Seattle Genetics : Research Funding. Gisselbrecht:Allos: Consultancy, Membership on an entity's Board of Directors or advisory committees.

Author notes

*

Asterisk with author names denotes non-ASH members.