Distribution of PD-L1, TROP2 and HER2- “lowness” in early triple-negative breast cancer: an opportunity for treatment de-escalation

Background HER2, TROP2 and PD-L1 are novel targets in triple-negative breast cancer (TNBC). The combined expression status of these targets, and whether they can define prognostic subgroups, is currently undefined. Methods Immunohistochemistry was used to determine HER2, TROP2 and PD-L1 levels in 459 TNBC cases, that received in the adjuvant/neoadjuvant setting active surveillance, CMF, anthracycline-, anthracycline plus taxane-, or carboplatin-containing regimes. Results HER2-low patients with PD-L1 > 1 CPS (double-positive, herein “DP”) had a mean PFS of 4768 days (95% CI: 4267–5268) versus 3522 days (95% CI: 3184–3861) for non-DP patients (P = 0.002). Regarding the received adjuvant treatment, DP patients (versus non-DP) receiving anthracyclines plus taxanes exhibited a mean PFS time of 4726 (95% CI: 4022–5430) versus 3302 (95% CI: 2818–3785) days (P = 0.039). Finally, 100% of DP patients that received a carboplatin-based regimen were long-term disease-free. Conclusions Early HER2-low, PD-L1-positive TNBC patients have a very good prognosis, particularly if treated with anthracycline/taxane- or carboplatin-containing regimes.


Introduction
Traditionally, triple-negative breast cancer (TNBC) has been viewed as a disease lacking druggable targets [1].Recently, the positive effects of immunotherapies in the early [2,3] and advanced [4][5][6] disease settings challenge this view.Similarly, the efficacy of the antibody-drug conjugates (ADCs) sacituzumab govitecan against TROP-2 and trastuzumab deruxtecan against HER2 have improved clinical outcomes in the general TNBC subpopulation [7] and in the HER2-low (positive, non-amplified) TNBC subpopulation [8], respectively.Whether these targets expression/coexpression and/or their levels of expression determine different TNBC sub-types, or drive its clinical history regardless of the use of those drugs is currently unknown; however, these datasets of the drug development focus on three novel targets (PD-L1, TROP2 and HER2) in this disease.
In the past, we have approached the complexity of TNBC with complex taxonomic techniques mixing genomics and phosphoproteomics to find prognostic [9] Maria Jose Bueno and Silvana Mouron have contributed equally to this manuscript.
or predictive [10] factors in the early disease setting.This previous work was performed relying on a growing collection of TNBC samples (174 cases in the first study and 218 in the second).Now, we have expanded this collection to 459 TNBC cases and adopted a more pragmatic approach: while multi-platform (geno-proteomics), multimarker-based classifications are unlikely to enter clinical routine practice soon due to technical, cost, and robustness or interpretation issues, histologic/immunohistochemistry markers are still the cornerstone of the routine decision tree in clinical routine breast cancer management.Since it is likely that these three targets are incorporated in the clinical routine soon, we wondered if the levels of these three new targets would condition the clinical course of TNBC cases not treated with drugs against them, and thus could be used to stratify TNBC cases to make clinical decisions; since our patient collection includes cases only up until 2015, none of them received (for the early disease setting management) any drug against those targets.The objective of this study was to determine the prognostic role of different combinations of PD-L1 and TROP-2 expression and HER2-low status in a cohort of early TNBC treated with chemotherapy only [cyclophosphamide plus methotrexate plus fluorouracil (CMF), anthracycline-, anthracycline plus taxane-, or carboplatin-containing regimes] or just active surveillance.Unexpectedly, we found that HER2-low, PD-L1-positive TNBC patients have an exceptional prognosis when treated with chemotherapy only, in particular when the treatment consisted on a carboplatin-based regime.

Methods
This cohort was constituted by consecutive TNBC cases diagnosed in 3 hospitals of the Madrid area between January 2005 and December 2012.Patients with lack of metastatic disease at diagnosis were candidates for this study.A tissue micro-array was mounted using the primary tumor paraffin blocks.PD-L1 was determined with the Combined Positive Score (CPS) index (antibody 22C3 from DAKO); an H-score (0-3) was calculated for TROP2 (ABS380 from ENZO), whereas HER2-low status was determined by immunohistochemistry (4B5 from Ventana).Clinical data were collected using medical records.Relapse-free rates and hazard ratios were calculated with the Kaplan-Meier and Cox's proportionate hazards methods, respectively.Two-sided P < 0.05 was considered significant for all tests.Data were analyzed with the SPSS Statistics Version 19 software.The study was approved by the Ethics Committee of the H12O (protocol approval #CEI 11/37); patients provided written informed consent.We followed the STROBE guidelines for cohort studies.
We then analyzed if the DP status on itself had prognostic (i.e., associated with better disease course in absence of treatment) or predictive (i.e., associated with benefit only in case of receiving certain chemotherapy regimes) value.To that end, we analyzed the PFS differences across the subgroups determined by the type of received adjuvant/neoadjuvant chemotherapy (or active surveillance; Table 1).Whereas for patients in the active surveillance group (Fig. 2d), CMF-treated (Fig. 2e) or anthracycline-treated (Fig. 2f) groups, we did not observe significant differences, DP status was associated with a more favorable disease course when patients received anthracyclines + taxanes or carboplatin-based regimes (Fig. 2g and h), with 100% of DP patients being relapsefree in the latter group.The relapse-free estimates for the Table 1 Clinical characteristics 1 AC, FEC or FAC 2 AC followed by weekly paclitaxel, FEC followed by weekly paclitaxel, or TAC 3   Determining TROP-2 levels did not refine risk stratification, neither when TROP-2 was considered alone or in combination with the positivity of HER2 and PD-L1 (Fig. 2i and j).

Discussion
The recent improvements in early [2,3] and advanced [4][5][6][7][8] TNBC are achieved at the expense of receiving up to 5-drug combos, which can cause significant permanent toxicity [11].Current trials testing the addition of ADCs may add a sixth one; of note, T-DXD or sacituzumab govitecan is associated with high rates of grade 3-4 toxic events [7,8], which may not be acceptable in the adjuvant setting.Whereas the determination of actionable targets is important for guiding drug development, it may be important as well for undertaking therapeutic decisions, particularly therapy de-escalation.
a TROP2 positivity was defined by showing an H-score in quartiles 1 to 3 of the H-score distribution (cut-off point H-score = 1.48) b Patients that were found to have an amplified HER2 were excluded from subsequent subgroup analysis + Double-positive patients: patients whose tumors were HER2-low and had a PD-L1 CPS > 1 #Average age was older for DP patients compared with non-DP patients, with borderline statistical significance (P = 0.059).Nodal invasion distribution was statistically significantly different in PD-L1 positive versus negative patients (P = 0.048).In addition, treatment distribution was different in DP patients, with a higher proportion of patients receiving an anthracycline plus taxane-or carboplatin-based treatment-regime (P = 0.017).Of note, all parameters were compared among subgroups with a Chi-squared test, and the comparisons were-non adjusted for multiple-comparison testing.Thus, although after adjustment the treatment or nodal invasion distribution might have not been statistically significantly different among groups, the more conservative approach highlighting this potential imbalance was adopted  Patients at risk (number censored) 78(0) 56( 11) 46( 7) 24(20) 11( 13) 5( 6) 0( 5) 203(0)131( 19) 98( 18) 51(41) 20(28) 4( 16) 0(4) Patients at risk (number censored) 183(0)128(20)104( 18)54( 46) 24(29) 8( 16) 0 (8) 74(0) 45( 5) 31( 4) 16( 12) 5( 10) 1(4) 0(1) Patients at risk (number censored) 59(0) 42( 10) 35( 6) 18( 16) 8(10) 3(5) 0(3) 220(0)143( 19)107( 19)56(44) 22(31) 5( 17) 0 (5) Patients at risk (number censored) 4(0) 2( 2) 0(1) 0(0) 33(0) 15( 6) 10( 1) 3 Patients at risk (number censored) 2(0) 2(0) 1(1) 1(0) 0(1) 29(0) 24( 1) 22(0) 17(2) 3( 12) 1(2) 0(1) Patients at risk (number censored) 12(0) 9(1) 9(0) 2( 6) 1(1) 1(0) 0(1) 46(0) 36(1) 28( 2) 16( 12) 8( 8) 2( 6 In this series, we report that early TNBC DP patients experience a favorable disease course when treated with chemotherapy alone.Of note, 100% of the patients are long-term disease free if treated with carboplatin-based regimens, an inexpensive option with predictable toxicity.Detecting a group with such a good long-term prognosis, that can be determine with inexpensive, easy and rapid techniques opens the opportunity to run de-escalation trials; clearly, the first external confirmation of our results should be in the Keynote 522 trial [2,3], where we could obtain a preliminary evaluation of whether immunotherapy adds any further therapeutic benefit over a carboplatin-based combo alone.This could be followed by an additional randomized de-escalation trial for this patient population which represents 26% of the TNBC cases. A question of key importance is which is the mechanism behind the increased sensitivity to platinum-based chemotherapy of the DP group, since understanding it may open avenues to turn sensitive as well the non-DP cases.First, observing the HR for disease relapse (Figs.2a, b and c), it seems that PD-L1 positivity is a greater contributor to the protective effect of the "DP status" (HR = 0.40, compared to HR = 0.48 for HER2lowness); however, interaction between the two factors should be explored in the future in independent, larger series.PD-L1 positivity has been correlated with TILs' infiltration [although their co-occurrence seems to exert complex interactions [12], which, in turn, has been related as well to improved disease course in response to neoadjuvant chemo-or chemo-immunotherapy [13]].However, in our series, we did not find that correlation, or a significant improvement in the prognosis accuracy of the model by adding TILs' percentage to it (data not shown).An alternative explanation is that PD-L1 positivity may be an indirect marker of homologous recombination deficiency [14][15][16][17] (which, on the one hand, given the alterations in DNA repair, it may be linked to an increased mutational burden and neo-antigenic load, followed by PD-L1 expression as a compensatory mechanism, and, on the other hand, it is per se a sensitization factor for platinum-based therapies).The contribution of HER2lowness is more intriguing.Although it has been reported that HER2-lowness is associated with a worse disease course in early hormone-positive breast cancer and a better disease course in TNBC [18], it has not been specifically associated with increased sensitivity to cisplatin.Future pre-clinical studies should address the interactions between PD-L1 and HER2 expression in DNA replication and/or its consequences in the STING-dependent immune response at the molecular level [19], as well as their impact in the tumor immune infiltrate.
The main limitation of our study is its retrospective nature and the relative low numbers in the subgroup analysis.The low numbers seem to affect, particularly, to the survival analysis in Fig. 2d and e, where the DP positive patients are less than 5 in both cases, an issue that does not happen, for example, in the analysis shown in 2 g or 2 h.
In conclusion, early HER2-low TNBC patients with a PD-L1 score of CPS > 1 have better prognosis compared to other TNBC patients, particularly when treated with carboplatin-or anthracycline plus taxane-containing regimes.HER2-low, PD-L1-positive status seems to be predictive of benefit from carboplatin-based treatment in the adjuvant setting.

Fig. 1
Fig. 1 Staining examples and disease course according to PD-L1 or HER2-low status.Examples of four patients with different combinations of the three tested markers: TROP2-negative or-positive, PD-L1 CPS below or above 1, and HER2-low or HER2-zero

Fig. 2
Fig. 2 Disease course by type of chemotherapy (or active surveillance) in HER2-Low, PD-L1-positive patients (DP patients).a Relapse-free survival of HER2-Low versus HER2-Zero TNBC patients.b Same comparison for patients with PD-L1 CPS above (positive) or below (negative) 1. c DP patients versus non-DP patients (i.e., PD-L1>1 and HER2-Zero, PD-L1<1 and HER2-Low, or PD-L1<1 and HER2-Zero).d to h Comparison of the relapse-free times of DP versus non-DP patients when they received no adjuvant/