Analytical and clinical validation of PATHWAY Anti-HER-2/neu (4B5) antibody to assess HER2-low status for trastuzumab deruxtecan treatment in breast cancer

In DESTINY-Breast04 (DB-04), safety and efficacy of HER2-targeted antibody-drug conjugate (ADC) trastuzumab deruxtecan (T-DXd) in previously treated HER2-low unresectable/metastatic breast cancer were established. This manuscript describes the analytical validation of PATHWAY Anti-HER2/neu (4B5) Rabbit Monoclonal Primary Antibody (PATHWAY HER2 (4B5)) to assess HER2-low status and its clinical performance in DB-04. Preanalytical processing and tissue staining parameters were evaluated to determine their impact on HER2 scoring. The recommended antibody staining procedure provided the optimal tumor staining, and deviations in cell conditioning and/or antibody incubation times resulted in unacceptable negative control staining and/or HER2-low status changes. Comparisons between antibody lots, kit lots, instruments, and day-to-day runs showed overall percent agreements (OPAs) exceeding 97.9%. Inter-laboratory reproducibility showed OPAs of ≥97.4% for all study endpoints. PATHWAY HER2 (4B5) was utilized in DB-04 for patient selection using 1340 tumor samples (59.0% metastatic, 40.7% primary, (0.3% missing data); 74.3% biopsy, 25.7% resection/excisions). Overall, 77.6% (823/1060) of samples were HER2-low by both central and local testing, with the level of concordance differing by sample region of origin and collection date. In DB-04, the efficacy of T-DXd over chemotherapy of physician’s choice was consistent, regardless of the characteristics of the sample used (primary or metastatic, archival, or newly collected, biopsy or excision/resection). These results demonstrate that PATHWAY HER2 (4B5) is precise and reproducible for scoring HER2-low status and can be used with multiple breast cancer sample types for reliably identifying patients whose tumors have HER2-low expression and are likely to derive clinical benefit from T-DXd. Supplementary Information The online version contains supplementary material available at 10.1007/s00428-023-03671-x.


Tissue Specimens for Preanalytical, Staining, and Analytical Validation Studies
Formalin-fixed paraffin-embedded (FFPE) breast cancer tumor specimens sourced from commercial vendors were used in analytical studies, except where indicated.All specimens were provided de-identified.Cases for analytical study inclusion were chosen based upon a HER2 IHC score assigned through pre-screening at Roche Tissue Diagnostics by a single pathologist or by consensus score (score assigned by two pathologists, plus a third pathologist when there were disagreements).

Preanalytical Tissue Processing Studies
The human breast cancer cell line, MDA-MB-361, which has demonstrated IHC 2+ HER2 expression [1], was used to generate xenograft tumors.To evaluate the impact of ischemia, xenograft tumors were harvested, kept at room temperature without solution until their respective time-point, and then fixed in 10% neutral-buffered formalin (NBF) for 24 hours before processing.The MDA-MB-361 xenograft tumors were sectioned, stained, and scored as indicated.For studies assessing different fixatives, xenograft tumors were harvested and placed into the indicated fixatives for 24 hours before processing.For tissue section thickness studies, FFPE blocks were sectioned from 2 to 10µM and stained as described.For cut slide stability, sections were stored at the indicated temperatures and humidity levels and stained at pre-determined time points to determine the impact of these parameters upon staining performance.

Immunohistochemistry
FFPE blocks were sectioned at 4 µm and mounted onto positively charged glass slides.
Except where noted, all slides were stored at 4°C in low relative humidity and stained within four months.All slides were stained on the VENTANA BenchMark ULTRA staining instrument (VENTANA Medical Systems, Inc.) equipped with VENTANA VSS Version 12.3 software.IHC staining was performed using the current FDA-approved, CE-marked recommended staining procedure for anti-HER2 IHC analysis (Supplemental Table 1), with the following staining protocol steps: The PATHWAY HER2 4-in-1 Control Slide stained with PATHWAY HER2 (4B5) antibody was used to ensure appropriate HER2 staining on each instrument run.FFPE tumor specimen-matched tissue specimens stained with the NRC were also reviewed for the presence and acceptability of nonspecific background staining.Once control slides were deemed acceptable, the PATHWAY HER2 (4B5) antibody-stained test slides were evaluated.

Staining Study with Variable Cell Conditioning and Antibody Incubation Times
Variances in the IHC staining protocol, particularly in the incubation time of the cell conditioning and antibody incubation steps, can significantly impact staining intensity [2].To evaluate whether IHC score and/or HER2-low status is affected by using alternative assay staining protocols, 13 FFPE resections (including 11 borderline cases near the 10% cutoff between IHC 0 and 1+) were selected and stained using the recommended staining protocol versus alternative protocols with variations in the cell conditioning and/or antibody incubation staining steps.The recommended incubation times for cell conditioning and antibody incubation are 36 minutes and 12 minutes, respectively.Reference slides from each FFPE tumor specimen were created using the recommended staining protocol with the PATHWAY HER2 (4B5) antibody and NRC.Experimental slides for each specimen were generated using the alternative staining protocols with the PATHWAY HER2 (4B5) antibody and NRC.Slides were randomized, and stained sections were evaluated using the HER2 IHC scoring method described in the main Methods section.All slides (reference and experimental) were evaluated by pathologists blinded to the reference score and staining procedure.

Intra-and Inter-Reader Precision
For inclusion in the reader precision study, a consensus score for HER2-low status was established for each case by a separate group of two pathologists.The precision study encompassed assessments of inter-reader and intra-reader concordance.This study included 100 breast carcinoma specimens (83 resections and 17 core needle biopsies) spanning the range of HER2 expression (41 IHC 0, 25 IHC 1+, 25 IHC 2+, and 9 IHC 3+), including 11 borderline specimens (10 borderline specimens near the 10% cutoff between IHC 0 and 1+ and 1 between IHC 2+ and IHC 3+).These samples were distinct from those used in other analytical studies.Slides for each specimen were stained with the BenchMark ULTRA staining instrument using the PATHWAY HER2 (4B5) antibody (or NRC) and the current FDA, CE-marked recommended staining procedure.Slides were de-identified, randomized, and evaluated twice by three internal (Roche) pathologists, to give six complete data sets, with a minimum of a two-week washout period between the two reads.All Roche pathologists had significant experience in evaluating the HER2 4b5 assay in breast (>5 years).
For inter-reader precision, the HER2-low status determined by one pathologist was compared to that determined by another pathologist for each case and deemed concordant or discordant.
For the purposes of statistical analysis, cases with IHC 1+ or 2+ were classified as HER2-low and cases with IHC scores of 0 and 3+ were classified as non-HER2-low.Average negative agreement (ANA) and average positive agreement (APA) were calculated for each pair of readers and aggregated across reader pairs to determine the overall APA and ANA.For intrareader precision, the HER2-low status (HER-2-low or non-HER2-low) from each pathologist's first read of a case was compared with that from their second read of the case and deemed concordant or discordant.The APA and ANA were calculated for each reader and aggregated across readers to determine the overall APA and ANA.For APA and ANA in each pairwise comparison, two-sided 95% confidence intervals (CIs) were calculated using the transformed Wilson score method.For APA and ANA aggregated across pairwise comparisons, two-sided 95% CIs were calculated using the percentile bootstrap method.

Intermediate Precision and Repeatability
Intermediate precision and repeatability studies were performed between three PATHWAY HER2 (4B5) antibody reagent lots, between three ultraView Universal DAB detection kit lots, between three BenchMark ULTRA staining instruments, between three days, and within two different instrument runs.For the antibody lot, detection kit lot and instrument studies, replicate slides were sectioned from 24 unique breast carcinoma tissue specimens and used for staining and evaluation.For between-day precision and within-instrument run repeatability studies, replicate slides were prepared from the same 24 unique breast carcinoma tissue specimens, stained with one antibody reagent and detection kit lot, and run on one instrument across three non-consecutive days.Of the 24 specimens (six IHC 0, six IHC 1+, six IHC 2+, and six IHC 3+) included in the study, four were HER2-low borderline cases (IHC 0/1+ at the 10% cutoff).Samples were randomized and de-identified, and one pathologist read all slides.For the purposes of statistical analysis, cases with IHC 1+ or 2+ were classified as HER2-low, and cases with IHC scores of 0 and 3+ were classified as non-HER2-low.Positive percent agreement (PPA), negative percent agreement (NPA), and overall percent agreement (OPA) for HER2-low status (HER2-low versus non-HER2-low) were evaluated.Two-sided 95% CIs were calculated from 2 000 bootstrap replicates using the percentile bootstrap method.For observations of 100.0%, the 95% CIs were calculated using the Wilson score method.

Interlaboratory Reproducibility
Interlaboratory reproducibility studies were performed at three external laboratories.In total, 28 breast cancer tissue specimens (eleven IHC 0, seven IHC 1+, seven IHC 2+, and three IHC 3+) were selected; of which four were IHC 0/1+ borderline cases.Sites were provided with five sets of slides to stain; laboratory personnel-stained slides on five nonconsecutive days spanning over ≥20 days, with the first and last staining days spaced ≥20 days apart.
Slides were de-identified and randomized according to a randomization schedule; thus, slides were stained and read in differing orders.Slides were stained using the recommended PATHWAY HER2 (4B5) assay procedure.Two pathologists per site (i.e., six pathologists in total) read slides.For the purposes of statistical analysis, cases with IHC 1+ or 2+ were classified as HER2-low and cases with IHC scores of 0 and 3+ were classified as non-HER2low.The APA, ANA, PPA, NPA, and OPA for HER2 IHC scores and/or HER2-low statuses were evaluated.Two-sided 95% CIs were calculated using the percentile bootstrap method using 2 000 bootstrap replicates.For observations of 100.0%, the 95% CIs were calculated using the Wilson score method (for PPA, NPA, and OPA) or the transformed Wilson score method (APA and ANA) [3].For the interlab reproducibility study, each site had two pathologist readers.All readers are board certified anatomic/surgical pathologists with an average of 13 years of clinical practice (range 6-23 years). 1 of 6 readers was a specialist in breast pathology.Supplemental Table 1 HER2: human epidermal growth factor receptor 2; IHC: immunohistochemistry.Test samples were processed on a BenchMark ULTRA using the PATHWAY HER2 (4B5) antibody with the U PATHWAY HER2 (4B5) staining procedure, with the indicated adjustments in the cell conditioning and antibody incubation times.Samples were compared with a sample processed using the recommended PATHWAY HER2 (4B5) assay of cell conditioning for 36 minutes and antibody incubation time of 12 minutes (reference IHC sample).Cases that switch between IHC 1+ and IHC 2+ are designated as having potential change in HER2-low status (which would depend on the ISH status of the sample).

Supplemental Table 3. Agreement in HER2 IHC scores with a 4 µm reference for breast cancer tissues of different section thicknesses
In total, 100 breast cancer specimens (83 resections and 17 needle core biopsies) spanning the range of HER2 expression, including 11 borderline HER2-low positive/negative cases, were stained and read twice by three pathologists.A washout period of two weeks occurred between the two reads.Two-sided 95% CIs were calculated using the percentile bootstrap method from 2000 bootstrap replicates.