Analysis Unable to Achieve Concordance Among PD-L1 Assays in TNBC

December 16, 2019

Conference | San Antonio Breast Cancer Symposium

Mathematical modeling was unable to harmonize the VENTANA SP263 and Dako 22C3 PD-L1 assays with the FDA-approved VENTANA PD-L1 SP142 companion diagnostic for the selection of patients with advanced triple-negative breast cancer appropriate for atezolizumab plus nab-paclitaxel.

Hope Rugo, MD

Accordingly, lead study author Hope S. Rugo, MD, and coinvestigators wrote in the conclusion of their poster at the 2019 San Antonio Breast Cancer Symposium (SABCS) that “the assays

cannot be considered equivalent,” leaving the SP142 assay as “the only clinically validated companion assay to select patients with metastatic TNBC for treatment with atezolizumab plus nab-paclitaxel.”

IMpassion130 was the first phase III trial to demonstrate clinical benefit of immunotherapy in patients with PD-L1—positive metastatic TNBC. Based on this study, the combination of atezolizumab and nab-paclitaxel was approved in the United States and Europe for this indication.

PD-L1 status in IMpassion130 was determined using SP142. Testing predicted progression-free survival (PFS) and overall survival (OS) with the combination of atezolizumab and nab-paclitaxel compared with nab-paclitaxel plus placebo. The absolute improvement in median OS in the PD-L1-positive population was 7 months (HR, 0.71; 95% CI, 0.54-0.94).²

SP142 is clinically validated and has been approved by the FDA for use as a companion diagnostic to identify the appropriate patients with metastatic triple-negative breast cancer (mTNBC) to receive the atezolizumab combination. The assay determines PD-L1 status as PD-L1 expression on immune cells (ICs) as a percentage of tumor area. As determined by SP142 testing, the PD-L1 expression cutoff was ≥1% for PD-L1—positive status in IMpassion130.

However, in several countries outside of the United States where atezolizumab plus nab-paclitaxel is approved for patients with metastatic TNBC and PD-L1 IC ≥1% status, there is no specific PD-L1 assay identified for use with this regimen. Further, despite the success with SP142 in IMpassion130, questions remain regarding the optimal approach for identifying which patients will benefit from the atezolizumab combination.

Accordingly, Rugo a professor in the Department of Medicine, Hematology/Oncology, and director, Breast Oncology and Clinical Trials Education at University of California, San Francisco, and coinvestigators conducted a posthoc analysis of the biomarker-evaluable population (BEP) from IMassion130, comparing SP142 with the 2 other PD-L1 assays: the complementary diagnostic VENTANA SP263 and the companion diagnostic Dako 22C3. These tests have been used for other immunotherapy agents in other tumor types. Results of this posthoc analysis were previously presented by Rugo et al at the 2019 ESMO Congress.³

There were 614 patients in the BEP, making up about 68% of the intent-to-treat (ITT) population. Baseline characteristics were well-balanced overall between the BEP and the ITT, except that there were more patients with PD-L1 IC—positive tumors per SP142 testing in the BEP at 46% versus 41%, respectively. The median PFS was 8.3 months with the atezolizumab regimen versus 4.1 months with nab-paclitaxel plus placebo in the PD-L1–positive BEP population. In the PD-L1–positive ITT population, the median PFS was 7.5 months versus 5.3 months, respectively.

To measure PD-L1 status in the BEP for their posthoc analysis, Rugo et al used the standard cutoffs for the 3 assays: IC 1% for SP142 and SP263, and combined positive score (CPS) 1 for 2263. The CPS is the total PD-L1 expression on both tumor cells and ICs.

Data presented at ESMO showed that 22C3 and SP263 were not concordant with SP142. The assays had what Rugo et al described as “subpar” overall percentage agreements (OPAs) with SP142 IC 1%, at 64% for 22C3 CPS 1 and 69% for SP263 IC 1%.

Demonstrating the nonequivalence of the assays, the PD-L1—positive populations identified by 22C3 and SP263 were larger than the SP142 population at 81% and 75% of the ITT, respectively. The SP142 population was a subgroup of these populations.

Rugo et al observed that the improved outcomes with the atezolizumab regimen in the 22C3 and SP263 populations were actually driven by the SP142 subgroup. Patients with the longest median PFS and OS and smallest hazard ratio estimates were identified by SP142. Also of note, little-to-no PFS and OS benefit was observed in patients who were PD-L1—positive by 22C3 or SP263, but PD-L1–negative by SP142.

At SABCS, Rugo et al presented a poster highlighting an additional exploratory analysis with 2 objectives: to evaluate whether using IC 1% as the cutoff with 22C3, instead of the standard CPS ≥1 used with this assay, would lead to analytical concordance with SP142 and enhance clinical utility; and identify alternative PD-L1 cutoffs for 22C3 or SP263 that would both analytically and clinically replicate the SP142 population.

With the first objective, Rugo et al did not find greater concordance with SP142 regarding the population identified or clinical outcomes when using the IC 1% cutoff with 22C3; the change in cutoff yielded similar outcomes as with the 22C3 standard cutoff of CPS 1, with an OPA now of 69%. SP142 IC was again a subgroup of the 22C3 population and again drove the clinical benefit observed.

Regarding the second objective of the exploratory analyses, mathematical modeling identified the optimal cutoff as CPS 10 for 22C3 and IC 4% for SP263. These were the cutoffs with the highest combined OPA, positive percentage agreement (PPA), and negative percentage agreement (NPA) between each assay and SP142.

For SP142 IC 1% and 22C3 CPS 10, the OPA, PPA, and NPA were all 74%. For SP142 IC 1% and SP263 IC 4%, the OPA, PPA, and NPA were 75%, 73%, and 77%, respectively.

“The clinical activity of atezolizumab plus nab-paclitaxel in patients identified by 22C3 CPS ≥10 and SP263 IC ≥4% yielded mixed results and was not equivalent to that observed in patients identified by SP142 IC ≥1%,” Rugo et al wrote.

The authors noted, however, that although analytical harmonization was not achieved, there was a similar PFS and OS benefit reached between the SP263 IC ≥4% subgroup and the SP142 IC ≥1% subgroup.

The median OS was 28.9 months in the atezolizumab arm of the SP263 IC ≥4% subgroup compared with 19.6 months with placebo (HR, 0.71). The corresponding OS figures were 27.3 months versus 17.9 months (HR, 0.74), respectively, in the in the SP142 IC ≥1% subgroup.

The median PFS in the SP263 IC ≥4% subgroup was 8.7 months for the atezolizumab regimen versus 5.5 months with nab-paclitaxel alone (HR, 0.64). The corresponding PFS figures were 8.3 versus 4.1 months (HR, 0.60), respectively, in the SP142 IC ≥1% subgroup.

Regarding the efficacy performance of the experimental SP263 cutoff, the authors noted, “Although SP263 IC ≥4% identified an additional population that may have PFS benefit, this cutoff also excluded 26% of SP142 IC ≥1% patients who may have derived PFS benefit from atezolizumab plus nab-paclitaxel.”

Unlike the SP263 cutoff, the experimental 22C3 cutoff did not reach a similar level of clinical benefit with the atezolizumab regimen as SP142 IC ≥1%.

The median OS was 22 months in the atezolizumab arm of the 22C3 CPS ≥10 subgroup compared with 18.7 months in nab-paclitaxel plus placebo arm (HR, 0.77). The median PFS was 7.5 months versus 5.5 months, respectively (HR, 0.71).

These median OS and PFS data, as well as the hazard ratios, are inferior to those derived with SP142 IC ≥1%.

References

Rugo HS, Loi S, Adams S, et al. Exploratory analytical harmonization of PD-L1 immunohistochemistry assays in advanced triple-negative breast cancer: A retrospective substudy of IMpassion130. Presented at: 2019 San Antonio Breast Cancer Symposium; December 10-14; San Antonio, TX. Abstract PD1-07.
Schmid P, Rugo HS, Adams S. Atezolizumab plus nab-paclitaxel as first-line treatment for unresectable, locally advanced or metastatic triple-negative breast cancer (IMpassion130): updated efficacy results from a randomised, double-blind, placebo-controlled, phase 3 trial [published online Nov 27, 2019]. Lancet Oncol. doi: 10.1016/S1470-2045(19)30689-8.
Rugo HS, Loi S, Adams S, et al. Performance of PD-L1 immunohistochemistry (IHC) assays in unresectable locally advanced or metastatic triple-negative breast cancer (mTNBC): Post-hoc analysis of IMpassion130. Presented at ESMO Congress 2019; September 27-October 1, 2019; Barcelona, Spain. Abstract LBA20.