Clinical evaluation of BCL-2/XL levels pre- and post- HER2-targeted therapy

Our previous pre-clinical work defined BCL-2 induction as a critical component of the adaptive response to lapatinib-mediated inhibition of HER2. To determine whether a similar BCL-2 upregulation occurs in lapatinib-treated patients, we evaluated gene expression within tumor biopsies, collected before and after lapatinib or trastuzumab treatment, from the TRIO-B-07 clinical trial (NCT#00769470). We detected BCL2 mRNA upregulation in both HER2+/ER- as well as HER2+/ER+ patient tumors treated with lapatinib or trastuzumab. To address whether mRNA expression correlated with protein expression, we evaluated pre- and post-treatment tumors for BCL-2 via immunohistochemistry. Despite BCL2 mRNA upregulation within HER2+/ER- tumors, BCL-2 protein levels were undetectable in most of the lapatinib- or trastuzumab-treated HER2+/ER- tumors. BCL-2 upregulation was evident within the majority of lapatinib-treated HER2+/ER+ tumors and was often coupled with increased ER expression and decreased proliferation. Comparable BCL-2 upregulation was not observed within the trastuzumab-treated HER2+/ER+ tumors. Together, these results provide clinical validation of the BCL-2 induction associated with the adaptive response to lapatinib and support evaluation of BCL-2 inhibitors within the context of lapatinib and other HER2-targeted receptor tyrosine kinase inhibitors.


Introduction
Adaptive responses to targeted-therapies compromise treatment effectiveness and confer drug resistant phenotypes [1]. Treatment-associated adaptive reprogramming often includes parallel upregulation of alternative receptors and/or activation of alternative pathways, which compensate for target inhibition and maintain tumor growth and survival [2][3][4][5][6][7][8]. Our previous work identified the upregulation of an anti-apoptotic program as an adaptive response to lapatinib-mediated inhibition of HER2 in vivo [9] and BEZ235-mediated inhibition of PI3K/ mTOR in vitro [10]. BCL-2 induction, mediated via FOXO-dependent transcription and capindependent translation, was a critical component of this pro-survival response. To determine whether BCL-2 upregulation could be identified in patients treated with HER2-targeted therapies, we utilized tumor samples that were previously collected during the TRIO-B-07 clinical trial (NCT#00769470). TRIO-B-07 was designed to include patients with HER2+/ER-or HER2+/ER+ breast cancers and biopsies before and after 2-3 weeks of either lapatinib and/or trastuzumab treatment (Fig 1). Prior to definitive surgery, patients were treated with additional neo-adjuvant DNA-damaging and anti-mitotic chemotherapies. To specifically address HER2-associated events, we focused on the pre-(baseline) and post-treatment (run-in) biopsies prior to chemotherapy-treatment. To specifically compare the effects of lapatinib versus trastuzumab, we excluded tumors treated with both agents and focused on tumors treated with single agents. As previously described [11], these tumor biopsies were profiled via microarrays (GSE130788) and prepared as FFPE slides.

Biospecimens
Tumor biopsies were collected from patients enrolled in the TRIO-B-07 clinical trial (NCT#00769470) as previously described [11]. TRIO-B-07 was reviewed and approved by multiple institutional review boards (UCLA; Olive View; Western), and all participants signed an IRB-approved informed consent form. The Institutional Review Board (IRB) of the Harvard University Faculty of Medicine reviewed and determined that our study is not human subjects research. The UCLA IRB also reviewed and approved our study. Our UCLA Data Use Agreement (DUA) confirmed that all samples were de-identified and that access to identifiers (or links to identities) of individuals from whom the samples were collected would not be granted under any circumstances. Tables 1-4 summarize patient-relevant characteristics corresponding to the subsets of samples analyzed.

RNA isolation and microarrays
RNA isolation and gene expression profiling (GSE130788) was performed as previously described [11]. Pre-treatment and post-treatment samples from the lapatinib and trastuzumab-treatment arms of GSE130788 were subjected to Significance Analysis of Microarrays (SAM) analyses using the siggenes (v1.62.0) package in R (v4.0.1). The log10 ratios from GEO for all probes of human BCL-2 homologous and BH3-containing proteins from the BCL-2 Database [12] were tested using the single class (for post-treatment comparisons) or 2-class (for ER+/ER-pre-treatment) d.stat method. For post-treatment, delta was 0.10, corresponding to an FDR of 3.8%. For pre-treatment, delta was 1.0, corresponding to an FDR of 2.6%. Heatmaps were generated with ComplexHeatmap (v2.4.3). For genes with multiple significant probes, the probe with highest variance is presented.

Microscopy
IHC images were digitized at the DF/HCC Tissue Microarray Core or the UCLA Translational Pathology Core on the Aperio Digital Pathology Slide Scanner (Leica Biosystems). The Aperio ImageScope software (Leica Biosystems) was used for image visualization and acquisition.

Statistics
Statistical analyses were performed using GraphPad Prism version 8 for MAC. For statistical tests, Wilcoxon matched-pairs signed rank test or Mann-Whitney test was applied. For correlation analyses, Spearman correlation was applied.
To address whether BCL2 mRNA expression correlated with BCL-2 protein expression, pre-and post-treatment FFPE tumors were evaluated for BCL-2 by IHC. We evaluated BCL-2 by intensity and proportion scores to calculate an H-score [13] on a case-by-case basis. Twenty-three lapatinib-treated and twenty trastuzumab-treated cases were evaluated by blinded pathological assessment (Tables 1-4). In contrast to HER2+/ER+ tumors, pre-  treatment BCL-2 protein levels were undetectable in most of the HER2+/ER-tumors (S1 Fig).
Despite BCL2 mRNA upregulation within HER2+/ER-tumors, only 2 out of 12 lapatinibtreated cases ( Fig 3A) and 1 out of 10 trastuzumab-treated cases ( Fig 3B) displayed a minor increase in BCL-2 protein levels following treatment. Eleven matched lapatinib-treated cases ( Fig 4A) and ten matched trastuzumab-treated cases ( Fig 4B) were HER2+/ER+; all of these cases were BCL-2-positive before treatment (H-scores = 15-218). Post-lapatinib, significant BCL-2 upregulation was evident in 9 out of 11 HER2+/ER+ cases (Fig 4A). Examples of BCL-2 and HER2 immunostaining of pre-and or post-treatment tumors from three lapatinib-treated cases [#35; #54; #71] are presented in Fig 5. Post-trastuzumab, BCL-2 upregulation was observed in 6 out of 10 HER2+/ER+ cases; however, these pairwise comparisons were statistically insignificant (Fig 4B). To determine whether treatment altered the expression of BCL-X L , another pro-survival protein, we performed BCL-X L IHC on the lapatinib-treated and trastuzumab-treated HER2+/ER+ cases. BCL-X L was upregulated in 4 out of the 9 cases where BCL-2 was upregulated in response to lapatinib (Fig 4C). Interestingly, upregulation in the absence of BCL-2 upregulation was detected within the remaining 2 out of 11 lapatinib-treated HER2 +/ER+ tumors. For the trastuzumab-treated HER2+/ER+ cases, BCL-X L was upregulated in only 1 out of the 6 cases where BCL-2 was upregulated ( Fig 4D) and within two additional cases; however, one of these cases displayed only a minor increase in BCL-X L protein levels post-treatment. These results indicate that BCL-2 is upregulated within the majority of lapatinib-treated HER2+/ER + tumors and that BCL-2 protein is undetectable pre-and post-treatment in most HER2+/ERtumors.   Since BCL2 is a direct transcriptional target of ER [14], we performed ER IHC and examined whether lapatinib-associated BCL-2 induction paralleled ER upregulation. ER was upregulated in 6 out of the 9 HER2+/ER+ cases where BCL-2 was upregulated in response to lapatinib (Fig 4E and S2 Fig) and in 2 out of the 6 HER2+/ER+ cases where BCL-2 was upregulated in response to trastuzumab (Fig 4F). By comparison, ER upregulation was detected within 2 out of 12 HER2+/ER-lapatinib-treated cases ( Fig 3C) and 1 out of 10 HER2+/ERtrastuzumab-treated cases ( Fig 3D); however, only one of these cases displayed a minor increase in BCL-2. These results indicate that BCL-2 upregulation in lapatinib-treated HER2 +/ER+ tumors is often, but not always, coupled with increased ER expression.

Discussion
Here, we provide clinical validation of the BCL-2 induction associated with the adaptive response to lapatinib-mediated HER2 blockade. The upregulation of BCL2 mRNA occurred in both HER2+/ER-as well as HER2+/ER+ patient tumors treated with lapatinib or trastuzumab; however, BCL-2 protein levels pre-and post-treatment were undetectable in most HER2+/ERtumors. The upregulation of BCL-2 protein was evident within the majority of lapatinibtreated HER2+/ER+ tumors and was often coupled with increased ER expression and decreased proliferation. Comparable BCL-2 upregulation was not observed within the trastuzumab-treated HER2+/ER+ tumors. Together, these results support clinical evaluation of BCL-2 inhibitors within the context of lapatinib and the treatment of HER2+/ER+ breast cancers.
Our results are consistent with in vitro studies of ER-mediated lapatinib resistance mechanisms [15,16], which provided evidence for ER and BCL-2 induction within lapatinib-resistant models. Our results are further supported by similar lapatinib-focused analyses of ER and BCL-2 co-expression [17], which demonstrated parallel upregulation of ER and BCL-2 within two pre-clinical models of treatment resistant HER2+ breast cancer as well as a subset of lapatinib-treated HER2+ clinical samples. The relationship between lapatinib-mediated HER2 blockade and ER induction is consistent with evidence that pathways activated (MAPK/ERK or PI3K/AKT) downstream of HER2-signaling negatively regulate ER through both epigenetic and transcriptional mechanisms [15,[18][19][20][21][22][23][24]; whereas, the correlation between BCL-2 upregulation and ER upregulation is consistent with evidence that BCL-2 is a direct target of ER transcriptional regulation [14]. Since BCL-2 upregulation was also observed in a few tumors without ER upregulation, our results also suggest that there can be ER-independent regulation of BCL-2. These results are consistent with multiple previous reports demonstrating that antiestrogen treatment only decreased BCL-2 expression in a subset of tumors analyzed [25][26][27][28].
The correlation between BCL-2 upregulation and decreased proliferation is consistent with evidence that HER2 inhibition confers adaptive reprogramming. Since lapatinib and trastuzumab have differential mechanisms of action [29] and variable impacts on HER2-signaling blockade [30], the lapatinib-associated BCL-2 responses could be directly related to superior HER2 inhibition achieved with lapatinib versus trastuzumab.
Endpoint analysis of pathological responses in the TRIO-B-07 clinical trial indicated that lapatinib-treated HER2+/ER+ tumors had the lowest pathological complete response rates [11]. Since these tumors were subsequently treated with additional neo-adjuvant DNA-damaging and anti-mitotic chemotherapies, we are unable to address the extent to which BCL-2 induction clinically impacted the lapatinib-associated pathological responses observed within HER2+/ER+ tumors. Nonetheless, we would predict that tumor populations distinguished by elevated BCL-2 could contribute to residual and recurrent disease. These predictions are based upon our own pre-clinical in vivo data demonstrating BCL-2 induction in lapatinib-treated HER2+ residual tumors [9] as well as additional in vitro and in vivo studies demonstrating BCL-2 induction within lapatinib-resistant models [15][16][17].
BCL-2 dependencies have been described with the context of HER2-/ER+ breast cancers, which are enriched for BCL-2 expression [31][32][33]. Based upon these observations, pre-clinical studies have determined that selective targeting of BCL-2 via ABT-199/venetoclax [34] enhanced the effectiveness of ER-targeted treatment [35] and have motivated clinical investigations of venetoclax combined treatment strategies within HER2-/ER+ tumors [36]. Since our results demonstrated similar BCL-2 enrichment within the context of HER2+/ER+ breast cancers, we would predict that HER2-targeted and ER-targeted treatments of these tumor types could be enhanced via venetoclax.
The pro-survival functions of BCL-2, BCL-X L and MCL-1 have been described within various contexts of HER2-targeted therapies. Using cell-derived and patient-derived xenograft models of HER2+/ER-breast tumors, our previous pre-clinical work provides evidence that dual targeting of BCL-2 and BCL-X L , via ABT-737 [37] or ABT-263/navitoclax [38], enhanced the in vivo effectiveness of lapatinib [9] or T-DM1 [39]. Another report identified an increased BCL-2:BAX ratio within a trastuzumab-resistant in vitro model and demonstrated that dual targeting of BCL-2 and BCL-X L enhanced the effectiveness of trastuzumab [40]. Multiple preclinical studies have identified MCL-1 as a critical component of lapatinib-resistant phenotypes and demonstrated that blockade of MCL-1, via obatoclax [41] or S63845 [42], enhanced treatment effectiveness [43][44][45][46][47]. Although several studies have demonstrated trastuzumabassociated downregulation of MCL-1 [48] or BCL-2 [49], these studies also reported that further disruption or modulation of the pro-survival proteins enhanced treatment effectiveness. Together with our TRIO-B-07 clinical results, these studies support the evaluation of BCL-targeted agents within the context of HER2-targeted treatments.