Xinqidi,Your biological research partner
Xinqidi Biotech Co.,Ltd,Wuhan,China 2008-2021
R&D 13th year

RAB5A expression is a predictive biomarker for trastuzumab emtansine in breast cancer

Issuing time:2021-11-15 10:52

Abstract

HER2 is a predictive biomarker for HER2-targeted therapeutics. For antibody–drug conjugates (ADCs; e.g., trastuzumab emtansine (T-DM1)), HER2 is utilized as a transport gate for cytotoxic agents into the cell. ADC biomarkers may therefore be more complex, also reflecting the intracellular drug transport. Here we report on a positive correlation between the early endosome marker RAB5A and T-DM1 sensitivity in five HER2-positive cell lines. Correlation between RAB5A expression and T-DM1 sensitivity is confirmed in breast cancer patients treated with trastuzumab emtansine/pertuzumab in the I-SPY2 trial (NCT01042379), but not in the trastuzumab/paclitaxel control arm. The clinical correlation is further verified in patients from the KAMILLA trial (NCT01702571). In conclusion, our results suggest RAB5A as a predictive biomarker for T-DM1 response and outline proteins involved in endocytic trafficking as predictive biomarkers for ADCs.

Introduction

The increased focus on personalized medicine has together with our increasing knowledge in cancer biology revealed a great potential for the use of biomarkers in cancer treatment. A number of different biomarkers are already incorporated in clinical practice to predict patient survival, select an appropriate therapy, or monitor disease progression1. Predictive biomarkers enable careful selection of those patients most likely to benefit from a specific treatment, and hence, such knowledge is crucial in order to rationally exploit current and future high-cost targeted cancer therapeutics. HER2 (ERBB2) is a validated biomarker in breast cancer and HER2 gene amplification or protein overexpression is found in ~20% of newly diagnosed breast cancer patients2,3. HER2 is utilized as a predictive biomarker for treatment with HER2- targeting monoclonal antibodies (mAbs) (trastuzumab and pertuzumab) and tyrosine kinase inhibitors (TKIs) (lapatinib and afatinib)3. The pharmacological effects of HER2-targeted mAbs and TKIs are a direct consequence of drug–target interaction and include antibody-dependent cellular cytotoxicity (ADCC) (mAbs), HER2 downregulation, and inhibition of growth-promoting signaling4,5,6. The ability of HER2 to undergo receptor-mediated endocytosis also makes this transmembrane protein a candidate for the delivery of cytotoxic agents into the cancer cells. This has indeed been exemplified by the antibody–drug conjugate (ADC) trastuzumab emtansine (T-DM1)7,8which received FDA approval for treatment of metastatic breast cancer in 2013. T-DM1 consists of trastuzumab linked by a thioether (N-maleimidomethyl cyclohexane-1-carboxylate (MCC)) to the highly cytotoxic maytansine-derived drug, DM19. Upon administration, T-DM1 binds to HER2 and is taken into the cell by HER2-mediated endocytosis. Proteolytic degradation of the trastuzumab component within the endo/lysosomal pathway is postulated as the mechanism for cytosolic release of DM1 which subsequently induces microtubule destabilization and cell death10,11. T-DM1, therefore, induces a cytotoxic mechanism of action within the cell in addition to the pharmacological effects generated by its trastuzumab component.

The action mechanisms of T-DM1 are clearly more complex than that of HER2-targeting mAbs12,13 and TKIs and we have evaluated whether this is reflected in the biomarkers that can be used to predict drug response. Candidate biomarkers for T-DM1 efficacy have, until now, focused on HER2 and its downstream signaling in addition to HER314,15, and little is known about the impact of proteins involved in endocytosis, endocytic vesicle transport, and exocytosis.

In this work, we report on a correlation between T-DM1 treatment response and RAB5A expression level. Our results show a significant correlation between RAB5A expression and T-DM1 sensitivity in a cell line panel in vitro. These results are further confirmed in T-DM1 treated patients from two independent clinical trials. The present study suggests RAB5A as a predictive biomarker for T-DM1 response.

Results

The cellular efficacy of T-DM1 does not correlate to trastuzumab sensitivity

The antiproliferative effects of the HER2-targeted mAb trastuzumab and the intracellular acting HER2-targeted ADC T-DM1 were established in five HER2-positive cell lines. Subjecting the cells to a 72 h treatment with trastuzumab or T-DM1 revealed the SK-BR-3 and AU-565 cells as highly sensitive to both therapeutics, whereas the SKOV-3 cells, on the contrary, were found non-responsive to trastuzumab and exhibited low sensitivity to T-DM1 as demonstrated by a relatively high IC50 of 1.2 µg/ml (Fig. 1 and Table 1). The HCC1954 and MDA-MB-453 cells were both found low- to moderately sensitive to trastuzumab treatment, but responded differently to T-DM1 with the HCC1954 cells showing high sensitivity, whereas the MDA-MB-453 cells showed low sensitivity (Fig. 1 and Table 1). No clear connection was therefore found between trastuzumab and T-DM1 sensitivity among the five cell lines (Fig. 1 and Table 1).

Fig. 1: In vitro sensitivity to trastuzumab and T-DM1.
figure1

Relative viability (MTT) of SK-BR-3, SKOV-3, AU-565, HCC1954, and MDA-MB-453 following 72 h treatments with indicated drugs. The sigmoid curve fit model a/(1 + exp(−(x − x0)/b)) was used for T-DM1. Data points represent the observed values of three independent experiments, the bar represents the average and the error bars represent the SD(trastuzumab). Source data are provided as a Source Data file.

Table 1 Cellular sensitivity of trastuzumab and T-DM1.

The T-DM1 sensitivity correlates to HER2 expression in HER2-positive breast and ovarian cancer cell lines

Strong HER2 expression was documented in the five HER2-expressing cell lines used in the present study compared to the low expression in MDA-MB-231 reported as HER2 negative (Fig. 2A)16.

Fig. 2: HER2 and RAB GTPase expression in cell lines.

A Representative Western blot of HER2 and γ-tubulin expression in SK-BR-3, SKOV-3, HCC1954, AU-565, MDA-MB-453, and MDA-MB-231 cells (n = 3). B Quantification of the HER2 Western blots relative to those of γ-tubulin. Data points represent the values of three independent experiments, the bars represent the average and the error bars represent the SD of the mean. C Linear regression analysis curve of HER2 protein expression and T-DM1 sensitivity (1/IC50(T-DM1)). D Representative western blot (n = 3) of RAB4A, RAB5A, RAB11A, and γ-tubulin expression in SK-BR-3, SKOV-3, AU-565, HCC1954, and MDA-MB-453 cells. EG Quantification of the RAB4, RAB5, and RAB11 Western blots relative to those of γ-tubulin. Data points represent the values of three independent experiments, the bars represent the average and the error bars represent the SD of the mean. Source data are provided as a Source Data file.

HER2 expression is essential for T-DM1 toxicity. However, it was hypothesized that the level of expression may not necessarily correlate directly to drug sensitivity due to differences in drug processing (e.g., uptake, intracellular transport, and interaction with intracellular drug targets) between the cell lines. Quantification of HER2 expression in the five positive cell lines indicated AU-565 to have the highest expression level of HER2, closely followed by HCC1954 (Fig. 2A, B). A ~50% lower HER2 expression was found in the SK-BR-3 cells compared to AU-565 while SKOV-3 and MDA-MB-453 were identified as the cell lines with the lowest HER2 expression in the panel (0.29 and 0.17 relative expression, respectively) (Fig. 2A, B). The level of HER2 expression reported here is in agreement with other reports16,17. Furthermore, a linear relationship was found between HER2 expression and T-DM1 sensitivity among the cell lines, resulting in an R2 value of 0.840 (Fig. 2C).

HER2-expressing cell lines differ in their expression level of proteins involved in endocytic trafficking

As T-DM1 is dependent on internalization and intracellular trafficking in order to exert its intracellular mechanism of action, proteins essential for endocytosis and exocytosis were quantified in the cell line panel. These proteins included RAB5A (Fig. 2D, E), implicated in the delivery of cargo from the plasma membrane to early endosomes as well as endosome fusion, RAB4A (Fig. 2D, F), implicated in recycling from early endosomes, and RAB11A (Fig. 2D, G), involved in perinuclear recycling of endosomes and plasma membrane–Golgi traffic18,19. The expression level of these proteins among the cell lines showed large differences and no simple connection was found between the expression levels.

RAB5A protein expression is highly correlated to T-DM1 sensitivity

It was further assessed whether the investigated proteins involved in endocytosis or exocytosis (Fig. 2D–G) had an impact on T-DM1 sensitivity in the evaluated cell lines. The expression level of RAB5A varied highly between the cell lines (Fig. 2D, E). However, a strong linear correlation was found between T-DM1 toxicity and RAB5A expression among the cell lines (R2 = 0.934) (Fig. 3A). In contrast, no linear correlations were found between T-DM1 sensitivity and expression of RAB4A or RAB11A (Fig. 3B, C). The cellular T-DM1 sensitivity was also correlated to the expression level of HER2 and RAB5A together revealing an R2 = 0.962 (Fig. 3D) which is higher than obtained with the correlations of HER2 (R2 = 0.840) (Fig. 2C) and RAB5A (R2 = 0.934) (Fig. 3A) alone. Thus, T-DM1 sensitivity correlates to the expression of RAB5A in this panel of cell lines, while no linear correlation is found for RAB4A and RAB11A. The expression level of HER2 and RAB5A together serves as a better biomarker for cellular T-DM1 response as compared to either of the two proteins alone.

Fig. 3: T-DM1 and RAB GTPase in vitro correlations.
figure3

Linear regression analysis curves between RAB5A (A), RAB4A (B), and RAB11A (C) protein expression (average of N = 3 as presented in Fig. 2) and T-DM1 sensitivity (1/IC50(T-DM1)) (average of N = 3 as presented in Fig. 1 and Table 1) in the five cell lines. D The linear regression curve between HER2 (average of N = 3 as presented in Fig. 2) × RAB5A protein expression and T-DM1 response. Each data point represents one cell line. Source data are provided as a Source Data file.

RAB5 RNA expression correlates with T-DM1 sensitivity in the I-SPY2 clinical trial

We then validated our in vitro finding in the I-SPY2 clinical trial by testing whether the clinical response toward T-DM1, as measured by pCR, could be correlated to RAB5A RNA expression. Figure 4A summarizes the pCR and the hormone receptor (HR) status of the T-DM1 + pertuzumab (T-DM1 + P) and trastuzumab and paclitaxel (TH) control arms. Overall, 30 of the 52 patients on the T-DM1 + P arm and 8 of the 31 patients in the control arm achieved a pCR. Normalized, pre-treatment expression levels of RAB4A, RAB5A, and RAB11A were tested individually as biomarkers of response. Of the biomarkers evaluated, only RAB5A was associated with response in the T-DM1 + pertuzumab arm (Fig. 4B and Supplementary Table 1, p = 0.01, LR test). None of the tested biomarkers were associated with response in the control arm (trastuzumab + paclitaxel) (Fig. 4B). The p-value for the interaction between RAB5A expression and treatment was 0.02 and remained <0.05 after adjusting for hormone status (Fig. 4B and Supplementary Table 1).

Fig. 4: T-DM1 and RAB GTPase correlations in the I-SPY2 cohort.
figure4

A Number of patients, hormone receptor (HR) status, and pathological complete responses (pCR) in the T-DM1 + pertuzumab and trastuzumab-paclitaxel arm of the I-SPY2 study. B Association plot summarizing qualifying biomarker analyses of RAB4A, RAB5A, and RAB11A expression levels as specific predictors of pCR to indicated treatment. Results are organized by the logistic model/data used along the rows, and the biomarker evaluated along the columns. Circle sizes are proportional to the significance (−log10 (LR test p)); and circle color reflects the magnitude of coefficient (red: positive, blue: negative) from each corresponding logistic model. White background indicates p < 0.05, and the odds ratio associated with 1 standard deviation increase in expression were also shown (in white) inside the circle. The 95% confidence intervals for the coefficients are found in Supplementary Table 1. Source data are provided as a Source Data file.

The association between RAB5A and T-DM1 + pertuzumab response may be attributable to pertuzumab rather than T-DM1. Although we cannot directly compare the two experimental arms, analysis of the trastuzumab + paclitaxel + pertuzumab group (n = 44) showed no significant correlation between RAB5A and pCR (Coef = 0.38 (95% CI: −0.36–1.17), LRp = 0.32) suggesting pertuzumab is not the key contributor to the significant pCR and RAB5A association for T-DM1 + pertuzumab-treated patients (Fig. 4B).

RAB5A is a constitutively expressed protein, and utilization as a biomarker will depend on the establishment of a threshold that separates RAB5Ahigh and RAB5Alow expressing cancer. Figure 5A shows boxplots of RAB5A expression levels stratified by arm and pCR status; based on a Monte-Carlo 2-fold cross-validation procedure as described in M&M, a normalized RAB5A RNA expression level of 9.76 was selected as a threshold (dotted gray line). The patient stratification into RAB5Ahigh and RAB5Alow groups in each arm is illustrated in Fig. 5Btogether with the pCR data.

Fig. 5: Patient distribution within pCR and RAB5A expression in I-SPY2.
figure5

A Boxplots of RAB5A expression levels stratified by arm and pCR status (N = 83). The midline represents the median of RAB5 expression levels within each group (T-DM1 + pertuzumab-treated, no pCR: N = 22 independent samples; T-DM1 + pertuzumab-treated, pCR: N = 30 independent samples; trastuzumab + paclitaxel-treated, no pCR: N = 23 independent samples; trastuzumab + paclitaxel-treated, pCR: N = 8 independent samples). The upper and lower limits of the box correspond to the 1st and 3rd quartile of RAB5A expression, respectively, with whiskers extending to 1.5 times the interquartile range from top/bottom of the box. Dots represent expression values for each individual; and color reflects subtype (orange: HR+; green: HR−). B Mosaic plot showing patient distribution within the RAB5A RNA-high and -low based on the threshold of 9.76 by arm and pCR status. CE The bayesian estimated pCR rates within the two treatment groups overall (C) as well as when divided in RAB5A RNA-high (D) and low (E) subsets. F, G ROC curves of the performance of RAB5A RNA as a biomarker in T-DM1 + pertuzumab-treated (F)- and trastuzumab + paclitaxel-treated (G) patients. Source data are provided as a Source Data file.

Using this optimal threshold, only 2 patients in the TH arm (6%) had RAB5A levels <9.76 and are considered RAB5Alow, as opposed to 23% (12/52) of the T-DM1 + P arm (Fig. 5B). This may in part be attributed to the difference in RAB5A expression between the two arms, where the RAB5A levels are significantly higher in the trastuzumab+paclitaxel arm than the T-DM1 + pertuzumab arm. Nevertheless, within the T-DM1 + P arm, 68% (27/40) of RAB5Ahigh patients achieved a pCR, in contrast to 25% (3/12) of RAB5Alow patients (Fig. 5B). Of note, no significant differences in HR status distribution or pre-treatment ERBB2 expression levels are observed among the RAB5Ahigh and RAB5Alow T-DM1 + P treated patients (% HR−: 30% vs. 42%, Fisher exact test two-sided p = 0.49, and median ERBB2: 11.2 (SD: 0.85) vs. 10.6 (SD: 1.27), Wilcoxon rank-sum two-sided p = 0.08, respectively).

Bayesian logistic modeling was used to estimate the pCR probability distributions within the T-DM1 + P and TH control arms in the overall HER2+ population as well as within the (predicted sensitive) RAB5Ahigh and (predicted insensitive) RAB5Alow subsets (Fig. 5C–E). The Bayesian estimated pCR probability is 68% (95% PI: 54–81%) in the T-DM1 + pertuzumab arm relative to 24% (95% PI: 9–38%) in the control arm in the RAB5Ahigh patients. In contrast, the estimated pCR probability is 28% (95% PI: 4–51%) in the RAB5Alow subset in the T-DM1 + pertuzumab arm and 42% (95% PI: 11–72%) in the trastuzumab + paclitaxel arm. For comparison, using the same model, the estimated pCR probability of the entire HER2+group is 61% (95% PI: 48–74%) in the T-DM1 + pertuzumab arm and 27% (95% PI: 12–41%) in the trastuzumab + paclitaxel arm.

Finally, we also established ROC curves to find the most appropriate cut-off for RAB5A expression in the T-DM1 + pertuzumab arm. Using this method, the optimal RAB5A cutoff was found at 9.76, the same as the one identified using the Monte-Carlo procedure. The AUC was 0.69, again indicating RAB5A as a predictive biomarker in this cohort (Fig. 5F). A ROC curve was also calculated for the TH control arm. (Fig. 5G). In contrast to the T-DM1 + P arm, the low AUC of 0.36 (below 0.5) suggested low RAB5A expression as a potential predictor for trastuzumab + paclitaxel response, although the number of patients (n = 2) with RAB5Alow in this arm is too small to evaluate (Fig. 5G).

RAB5A protein expression correlates with T-DM1 sensitivity in a subset of patients in the Kamilla study

Our positive correlation between RAB5A expression and T-DM1 sensitivity both in vitro and in the I-SPY2 clinical cohort was further verified in a small subset of the KAMILLA study including 19 patients treated at Oslo University hospital with biobanked primary biopsies. A highly significant (p < 0.03) difference in PFS was found between patients with RAB5A Allred scores 7 and 8, where 7/8 patients had a PFS of 50 weeks or more, and Allred scores 0-6, where all patients had PFS less than 50 weeks (Fig. 6A). The Allred score and PFS data for each patient is included in the data source file and exemplified IHC images are shown in Fig. 6B.

Fig. 6: T-DM1 and RAB5A correlation in a subset of patients in KAMILLA.

A Box blot indicating the distribution of PFS in RAB5A low (Allred score 0–6, N = 11 independent samples) and RAB5A-high (Allred score 7 and 8, N = 8 independent samples) expressing patients. The midline represents the median of progression-free survival. The upper and lower limits of the box correspond to the 75% and 25% percentile of progression-free survival respectively; the whiskers represent the 90% and 10% percentile. The data points represent the individual data points in each group. A Mann–Whitney rank-sum test was used to determine the statistical difference in the median values between the two groups. B Exemplified images (×40 objective) of RAB5A IHC for Allred score 0 (patient #20), Allred score 3 (patient # 19), Allred score 6 (patient #21), Allred score 7 (patient #24) and Allred score 8 (patient #15). Scale bars: 100 µm. All samples were manually screened by 2 observers. Consensus on staining intensity and amount was obtained, and representative areas defined. Number of images captured from each slide: patient #20 4 images, patient #19 10 images, patient #21 8 images, patient #24 6 images, patient #15 4 images. Source data are provided as a Source Data file.

Discussion

Most of the targeting drugs currently approved for the treatment of cancer are mAbs or small-molecular inhibitors for which the drug target also represents the target for the mechanism of action. The target itself represents a clear biomarker for treatment with these drugs, although other factors may be important to identify patients likely to experience resistance or low tolerability. In complex targeting therapeutics incorporating a cytotoxic component such as in ADCs, other biomarkers associated with the intracellular transport and/or cytotoxic mechanism of action are likely to impact the therapeutic outcome12,13. In this study, the need for separate predictive biomarkers for ADC and mAbs treatment is emphasized by the lack of coherence between trastuzumab and T-DM1 sensitivity in the selected panel of HER2-positive cell lines. We demonstrate a linear correlation between cellular HER2 expression and response toward T-DM1 (Fig. 2C). This is in agreement with several clinical studies demonstrating higher response rates of T-DM1 in patients with HER2 mRNA levels above the median compared to the below-median subgroup14,15,20. Importantly, despite its highly targeted mechanism, T-DM1 shows clinical benefit only in a subset of HER2- positive breast cancer patients with an objective response rate reported to ~40%8,21,22. Furthermore, in HER2-positive gastric cancer, an objective T-DM1 response rate of only ~20% with no increase in efficacy compared to taxanes is reported23. The diverse clinical response toward T-DM1 in HER2-expressing cancer indicates drug efficacy to depend also on other factors than the extracellular target expression.

T-DM1 is dependent on endocytosis in order to transport its cytotoxic payload (emtansine) into the cell and we here present the first report on T-DM1 efficacy correlated to proteins involved in endocytosis and endocytic trafficking. Of the three proteins investigated (RAB4A, RAB5A, and RAB11A), only RAB5 expression was found to correlate with T-DM1 toxicity in a cell line panel of HER2-positive breast and ovarian cancer. The correlation of T-DM1 sensitivity and RAB5A protein expression was stronger than observed for HER2 expression, and the highest correlation was found when combining RAB5A and HER2 protein expression indicating incorporation of RAB5A together with HER2 as a better predictive biomarker for T-DM1 sensitivity. RAB5A is localized to early endosomes and regulates both endocytosis and endosome fusion of clathrin-coated vesicles24. The lack of correlation observed for RAB4A and RAB11A may reflect the intracellular processing of T-DM1 upon uptake. RAB4A-mediated recycling from early endosomes may, e.g., be limited25,26,27 and it is possible that DM1 escape the endocytic vesicles prior to accumulation in RAB11A-positive recycling endosomes. We confirmed the correlation between RAB5A expression and sensitivity to neoadjuvant T-DM1 in the I-SPY2 study. Even though the number of patients is small, our clinical data from I-SPY2 also illustrates the possibility to define a RAB5A expression threshold to determine T-DM1 treatment in HER2-expressing breast cancer patients. A subset of patients from the Kamilla study was here used as a verification cohort. As compared to the I-SPY2 cohort, the patients from the Kamilla cohort all had advanced progressive disease and had all previously been exposed to different treatment regimens, including chemo- and HER2-targeted therapies. Nevertheless, a highly significant correlation was found between RAB5A expression in primary tumor and PFS while on T-DM1 therapy. Altogether, our presented in vitro findings showing a significant correlation between T-DM1 treatment outcome and RAB5A expression, is here confirmed in two independent clinical cohorts.

It has previously been shown that RAB5A expression predicts poor prognosis of breast cancer patients28, and the current study strongly indicates that T-DM1 should be further evaluated as the treatment of choice for these patients.

T-DM1 is currently approved only for advanced HER2-positive breast cancer previously treated with trastuzumab, but a defined RAB5A threshold-biomarker holds promise also for patient stratification at earlier stages of this disease29, in addition to other indications, such as gastric cancer23.

Although the number of patients with RAB5Alow receiving trastuzumab in the I-SPY2 cohort is far too low to draw any conclusions, our results may also point toward a negative correlation between RAB5A expression and trastuzumab response. Thus, HER2-expressing cancers with low RAB5A do better on trastuzumab as compared to T-DM1. If low RAB5A can be used not only to deselect patients for T-DM1 but also to find those patients most likely to benefit from trastuzumab, it would be an improvement for future personalized HER2-positive breast cancer therapy. Altogether, better stratification of HER2-expressing breast cancer patients into treatment groups may add to the benefit of the treatment, and the deselected patients may also be offered alternative treatment at an earlier time point.

As the mechanism of T-DM1 action involves endocytosis, we hypothesized that proteins involved in the endocytic process had an impact on the treatment response to T-DM1. Our hypothesis was here confirmed first in vitro and then in two independent clinical cohorts. Furthermore, we believe our results also demonstrate a more general concept in which proteins involved in endocytosis and/or endocytic trafficking are utilized as biomarkers for ADCs. Even though RAB5A was the only candidate to succeed as a predictive biomarker for T-DM1 in the present study, this may be different for other ADCs dependent on both their targeting moiety and cytotoxic payload. Overall, our results imply that current and future ADC development and treatment will benefit from the incorporation of biomarkers reflecting uptake and intracellular transport.


Article classification: Biological abstract
Share to:
Tel:+86-027-87610298
Tel:+86-027-87610297
Add:Room A11-329, 1st Floor, No.1, SBI Venture Street, Optics Valley, East Lake
New Technology Development Zone, Wuhan, China.
Certificate NO.:U18Q28010569R0S