Purpose: Treatment of BRAF-mutated melanoma tumors with BRAF inhibitor–based therapy produces high response rates, but of limited duration in the vast majority of patients. Published investigations of resistance mechanisms suggest numerous examples of tumor adaptation and signal transduction bypass mechanisms, but without insight into biomarkers that would predict which mechanism will predominate. Monitoring phenotypic response of multiple adaptive mechanisms simultaneously within the same tumor as it adapts during treatment has been elusive.

Experimental Design: This study reports on a method to provide a more complete understanding of adaptive tumor responses. We simultaneously measured in vivo antitumor activity of 12 classes of inhibitors, which are suspected of enabling adaptive escape mechanisms, at various time points during systemic BRAF inhibition. We used implantable microdevices to release multiple compounds into distinct regions of a tumor to measure the efficacy of each compound independently and repeated these measurements as tumors progressed on systemic BRAF treatment.

Results: We observed varying phenotypic responses to specific inhibitors before, during, and after prolonged systemic treatment with BRAF inhibitors. Our results specifically identify PI3K, PDGFR, EGFR, and HDAC inhibitors as becoming significantly more efficacious during systemic BRAF inhibition. The sensitivity to other targeted inhibitors remained mostly unchanged, whereas local incremental sensitivity to PLX4720 declined sharply.

Conclusions: These findings suggest redundancy of several resistance mechanisms and may help identify optimal constituents of more effective combination therapy in BRAF-mutant melanoma. They also represent a new paradigm for dynamic measurement of adaptive signaling mechanisms within the same tumor during therapy. Clin Cancer Res; 22(24); 6031–8. ©2016 AACR.

Translational Relevance

Cancer is increasingly viewed as an adaptive disease in which response to drugs can vary as tumors evolve during therapy and over time. In BRAF-mutant melanoma, many potential adaptive signaling mechanisms have been reported, which are suspected to contribute to widespread disease recurrence. We demonstrate repeated use of microdevices in live tumors to measure in situ phenotypic response to many potential escape pathways in parallel. In clinical practice, these microdevices may be placed inside the tumor in a minimally invasive manner and provide parallel phenotypic readouts of tumor vulnerability to additional targeted or cytotoxic inhibitors at various time points during standard-of-care treatment. This may provide a new personalized treatment paradigm in which physicians rationally manage drug therapy by identifying optimal combinations throughout the treatment cycle. Importantly, the microdevices provide actionable phenotypic drug readouts from within the tumor microenvironment and may expand clinical biomarker-based selection beyond genomic testing.

Mutations in the BRAF gene occur with significant frequency in melanoma and several other cancers and have been targeted successfully by multiple drugs in the clinical setting. High initial response rates are achieved by BRAF inhibition in melanoma (1), but a majority of patients relapse within 9 to 12 months with more aggressive tumors that are increasingly resistant to a range of other therapies (2).

Tumor recurrence is thought to be the result of adaptive mechanisms by which tumor cells respond to inhibition of their preferred oncogenic signaling pathway with an upregulation of alternative pathways for survival and proliferation (3, 4).

In BRAF-mutated melanomas, significant adaptive responses (and thus potential targets for therapeutic combinations) have been reported, for instance, involving MEK inhibitors (3, 5). The clinical implications are far reaching. Although systemic treatment with one agent can significantly affect the response to other potential subsequent therapies, such systematic, parallel investigations of multiple agents or combinations are usually performed only for small numbers of compounds in vivo due primarily to the rapidly increasing cohort size required to assess combinations between multiple agents. Insights are often restricted to in vitro studies, but tumor responses to drugs are not only determined by an individual tumor's genome, but also by the complex interaction of tumor cells with their microenvironment, including immune and stromal cells, and many known and yet unknown factors that can dramatically alter phenotypic drug response (6–8).

We demonstrate in this study parallel assessment of phenotypic in situ drug response to inhibitors of the majority of signaling pathways known to be relevant and druggable in this cancer type. The approach uses an intratumor implantable microdevice for the simultaneous delivery of 18 drug compounds into isolated nonoverlapping regions of tumor (adapted from ref. 9), including targeted inhibitors of BRAF, Erk, CDK4/6, PI3K, EGFR, C-Met, MDM2, platelet-derived growth factor receptor (PDGFR), FGFR1, histone deacetylase (HDAC), and HSP-90. Through appropriate spacing and sizing of reservoirs and formulation of drug compounds, it can be ensured that the drug contents from a given reservoir do not disseminate to the vicinity of an adjacent reservoir within the time course of the experiment (9). The antitumor effect of the compounds is assessed by targeting each of these signaling nodes at multiple treatment time points in a given tumor: before, during, and after systemic inhibition of BRAF. Our results show great diversity in how targeted BRAF inhibition affected the intratumor response to the various agents very differently. Although the response to many agents remained virtually constant, prolonged BRAF inhibition induced a significantly increased sensitivity of tumors to agents targeting PI3K, PDGFR, EGFR, and HDAC.

Study design

The objective of the studies in Figs. 2 and 3 is to show biologic response to release of drugs and to test whether this response was significantly different between different treatment time points. Sample sizes were chosen to demonstrate statistical significance by Student t test between biologically distinct conditions or outcomes. Tissue sections were scored by an ImageJ image analysis algorithm in a blinded manner (see below). Only biologic replicates were used in data analysis. Average values and SDs are from 8 samples for all studies. Data from tissue sections were only excluded in the rare event that the tissue section was damaged during retrieval or was found to be entirely necrotic by IHC.

Device preparation and implantation

Microdose drug delivery devices were manufactured as described in the study by Jonas and colleagues (9). In short, cylindrical, microscale devices with 0.82 mm (diameter) × 4 mm (length) were manufactured from medical-grade Delrin acetal resin blocks (DuPont) by micromachining (CNC Micro Machining Center). Circular reservoirs (8–12 per device) were shaped on the outer surface of devices in dimensions of 200 μm (diameter) × 250 μm (depth). Drug–polymer mixtures were packed into device reservoirs using a tapered, metal needle (Electron Microscopy Sciences) until the reservoirs were completely filled. Each drug reservoir in the microdevice contained approximately 1 μg of the compound.

Devices were implanted directly into the mouse tumor using a 19-gauge spinal biopsy needle (Angiotech) and a retractable needle obturator to push the device into the tissue. Devices containing the drug remained in situ for 24 hours. All drug compounds were purchased from Selleck Chemicals, Inc.

The flank tumor was excised and the tissue containing the device was fixed for 24 hours in 10% formalin and perfused with paraffin. This specimen was sectioned using a standard microtome, and tissue sections were collected from each reservoir. Sections were antibody stained by standard IHC using cleaved caspase-3 antibody (Cell Signaling Technology) and scored using an ImageJ image (v1.48) analysis algorithm in a blinded manner.

Scoring of IHC sections

Percent area of DAB-stained cells was calculated for 400 μm × 400 μm regions in tissue directly perpendicular to the reservoirs. The analysis was performed in a semiautomated fashion using ImageJ Color Deconvolution with the vector “H DAB.” A threshold was set to eliminate background staining, and all outliers smaller than 2 pixels in diameter were removed. The Smooth function was also used to fill in cells that did not have completely homogenous staining. Error bars represent one SD. In a small number of samples (<10%), device reservoir was located in a region of tumor that was highly necrotic. These sections were not included in the analysis because the differential effect caused by the presence of drug in this tissue region cannot be determined independently.

Statistical analysis

Apoptotic index (AI) was calculated as (i) AI = % cleaved caspase-3–positive cells/total cells within 400 μm from reservoir–tissue interface for device sections; and (ii) AI = % cleaved caspase-3–positive cells/total cells for entire tumor section for systemic studies. For graph 3, curves shown are averaged over 8 samples for each time point. All error bars represent one SD. Coefficient of variation is defined as SD divided by mean. Significance tests and P values were calculated using Student t test, using normal-based 95% confidence intervals. Two-sided testing was used. The datasets were verified to be normally distributed.

Cell lines, mice, xenograft tumors, and systemic dosing

Human cancer cell lines A375 were passaged in DMEM with 10% FBS, 1% pen–strep (10,000 U/mL penicillin and 10,000 μg/mL streptomycin), and 1% of 200 mmol/L l-glutamine. All supplies were purchased from Life Technologies. Cells were maintained in conditions of 37°C and 5% CO2 and passaged 1 to 3 times per week on reaching approximately 70% confluence. Tumor cells (∼3 × 106) in 150 μL PBS were injected subcutaneously into the flanks above the hind limbs of 6- to 8-week-old female athymic nude mice (Crl:NU(NCr)-Foxn1nu) purchased from Charles River Laboratories. Tumors grew for approximately 2 weeks, to a size of 5 to 9 mm, before the first device implantation. Animals were chosen randomly for experiments when tumor size was between 5 and 9 mm. No specific method was chosen for randomization.

Tumor-bearing mice were dosed daily for 15 days with PLX4720 at 45 mg/kg by oral gavage. At 3 distinct time points (before treatment began, day 0; during early treatment, day 4; and after treatment, day 16), microdevices were implanted into the tumors and remained in situ 24 hours. After the final device retrieval, mice were euthanized to collect the tumors. In some cases, mouse tumors were biopsied with an 11-G Cassi biopsy gun (Scion) to retrieve the tissue immediately surrounding the device. All animal studies were conducted in accordance with protocols approved by the Committee on Animal Care at the Massachusetts Institute of Technology (Cambridge, MA; protocol 0415-038-18).

In vitro survival assay

For cell growth assays, A375 cells were treated with various concentrations of PLX4720 (0.1 or 1 μmol/L) for 7 days. Then, 10,000 cells were plated per well in a 96-well plate and treated with erlotinib, vorinostat, imatinib, or BYL719 (at 0.8, 4, or 20 μmol/L). Cells were counted at 24 hours according to a resazurin assay (PrestoBlue; Life Technologies) using the manufacturer's instructions.

Western blot analysis

Standard procedures were used for protein electrophoresis and Western blotting. A375 cells were treated with 1 μmol/L PLX4720 for 7 days and then with a second drug for 24 hours (erlotinib, imatinib, vorinostat, and BYL719 at 2 μmol/L). Cells were then lysed in 25 mmol/L Tris, 150 mmol/L NaCl, 10% glycerol, 1% NP 40, and 0.5 M EDTA with a protease Mini-complete protease inhibitors (Roche) and a phosphatase inhibitor cocktail (PhosSTOP, Roche) at 4°C. Protein lysates were separated by SDS-PAGE, transferred to a nitrocellulose membrane, blocked with Odyssey Blocking Buffer (LI-COR), and incubated in primary antibody overnight at 4°C. Proteins were detected using LI-COR secondary antibodies. Protein level intensity was measured with ImageJ, and data were pooled from at least three different experiments. Antibodies used were pAktS473 (CST #9271), pRb S807/811 (#8516, CST), pERK (E4) sc-7383 (Santa Cruz Biotechnology), and GAPDH (Sigma, G9545).

Animals harboring A375 tumors were dosed systemically for 15 days with the BRAF inhibitor PLX4720. At three distinct time points (before treatment began, day 0; during early treatment, day 4; and after treatment, day 16), microdevices were implanted into the tumors and remained in situ 24 hours. These microdevices carried spatially distinct reservoirs, loaded with doxorubicin, PLX4720, erlotinib, crizotinib, BYL719, lapatinib, palbociclib, imatinib, 17-DMAG, vorinostat, PD173074, YH239-EE, and SCH772984. Drugs were released from reservoirs during the incubation time into confined regions of tumor tissue and interacted with the local tissue in its native microenvironment. The devices were then excised with surrounding tissue in a minimally invasive manner and analyzed. Phenotypic responses for each drug were obtained in this manner before BRAF inhibition, during initial treatment regimen of 4 days, and after a total of 15 consecutive daily treatments with PLX4720. The experimental approach is shown schematically in Fig. 1. The 13 drugs tested on each microdevice are listed in Table 1 with their respective molecular targets. The microdevice provides a unique way to test drug sensitivity of a tumor within its native microenvironment and without systemic exposure or toxicities. It is possible to examine the effect of multiple drugs independent of one another and all within a single tumor because drug microdoses are only released over a small and confined region of tumor. Formulation of therapies in a (poly)ethylene-glycol polymer matrix ensures that diffusion distances of drugs are limited to 300 to 400 μm within 24 hours (9), which is significantly lower than the separation distances between nearest adjacent reservoirs of 900 μm.

Figure 1.

A, Schematic of tumor growth versus days of systemic treatment with a BRAF inhibitor. The three arrows show when microdevice tests were performed on the tumor: before treatment began; during early treatment; and at the end of the 15-day systemic treatment. B, Concept for in vivo drug sensitivity assay: device is implanted directly into tissue where drugs diffuse into confined regions of tumor. Each region is assayed independently to assess the tumor-specific response to a given drug. Following incubation, a second biopsy procedure is administered in which a coring needle selectively retrieves a small column of tissue that immediately surrounds the device. This tissue contains the regions of drug diffusion and is sufficient for determination of efficacy of drugs.

Figure 1.

A, Schematic of tumor growth versus days of systemic treatment with a BRAF inhibitor. The three arrows show when microdevice tests were performed on the tumor: before treatment began; during early treatment; and at the end of the 15-day systemic treatment. B, Concept for in vivo drug sensitivity assay: device is implanted directly into tissue where drugs diffuse into confined regions of tumor. Each region is assayed independently to assess the tumor-specific response to a given drug. Following incubation, a second biopsy procedure is administered in which a coring needle selectively retrieves a small column of tissue that immediately surrounds the device. This tissue contains the regions of drug diffusion and is sufficient for determination of efficacy of drugs.

Close modal
Table 1.

List of drugs tested during systemic BRAF inhibition with respective molecular targets

CompoundPLX4720Byl719ImatinibCrizotinib17-DMAGErlotinibLapatinib
Molecular target BRAF V600E PI3K V-Abl, c-Kit, PDGFR C-Met, ALK HSP90 EGFR HER2/EGFR 
Compound Palbociclib PD173074 Vorinostat YH239-EE SCH772984 Doxorubicin  
Molecular target CDK 4/6 FGFR1 HDAC p53-MDM2 ERK1/2 DNA-damaging agent  
CompoundPLX4720Byl719ImatinibCrizotinib17-DMAGErlotinibLapatinib
Molecular target BRAF V600E PI3K V-Abl, c-Kit, PDGFR C-Met, ALK HSP90 EGFR HER2/EGFR 
Compound Palbociclib PD173074 Vorinostat YH239-EE SCH772984 Doxorubicin  
Molecular target CDK 4/6 FGFR1 HDAC p53-MDM2 ERK1/2 DNA-damaging agent  

Abbreviation: ALK, anaplastic lymphoma kinase.

By performing biopsies of the device with surrounding tissue, we assay a single tumor's response to many drug compounds individually at multiple time points, without the need for whole-tumor resection. Device/tissue specimens are sectioned layer-by-layer, and analysis is performed on tissue sections that contain regions of tumor that are exposed to only one drug. Immunohistochemical staining provides a phenotypic readout of each drug's activity in the tumor. The preferred metric of a drug's local intratumor efficacy is the AI, defined as the percentage of apoptotic cells versus total cells within the exposed tumor region.

Initial drug sensitivities

Prior to any systemic treatments, we implanted the microdevices into tumors and tested initial sensitivities to all agents listed in Table 1. These melanoma tumors are initially highly sensitive to local microdose treatment with PLX4720, as exhibited by an AI of 40%. This sensitivity is likely due to the BRAF V600E mutation that is reported for this model (10, 11). We also determined the tumors to be highly sensitive to doxorubicin (AI = 52%) in concordance with previously reported results, as well as in vitro and whole-animal studies (9, 10).

The a priori sensitivities to all other targeted agents listed in Table 1 were comparatively low. These ranged from AI = 8% to 22%, with HDAC, MDM2, and MEK1/2 inhibitors among the more potent agents (Figs. 2 and 3).

Figure 2.

Representative tumor sections for each drug and time point tested via the microdevice. Tumor sections are stained for cleaved caspase-3 (brown cells) indicating apoptosis. The ratio of apoptotic versus all cells (AI) is used as a measure of drug efficacy.

Figure 2.

Representative tumor sections for each drug and time point tested via the microdevice. Tumor sections are stained for cleaved caspase-3 (brown cells) indicating apoptosis. The ratio of apoptotic versus all cells (AI) is used as a measure of drug efficacy.

Close modal
Figure 3.

A, Quantitative analysis for each drug, showing changes in intratumor treatment response versus PLX4720 systemic treatment time. For each drug, its apoptotic response is shown at three time points: day 0, before systemic PLX treatment has begun; day 4, after 4 days of systemic treatment; day 16, one day after the 15th and final day of systemic treatment. N = 8 for each time point and drug. Error bars, one SD. B, Ranking of absolute changes in AI for each of the drugs tested on the microdevice during systemic BRAF inhibition from day 0 to day 16.

Figure 3.

A, Quantitative analysis for each drug, showing changes in intratumor treatment response versus PLX4720 systemic treatment time. For each drug, its apoptotic response is shown at three time points: day 0, before systemic PLX treatment has begun; day 4, after 4 days of systemic treatment; day 16, one day after the 15th and final day of systemic treatment. N = 8 for each time point and drug. Error bars, one SD. B, Ranking of absolute changes in AI for each of the drugs tested on the microdevice during systemic BRAF inhibition from day 0 to day 16.

Close modal

Rapid changes during initial phase of systemic BRAF inhibition

Local microdose effect of PLX4720 decreases significantly (from AI = 40%–24%) on day 4 of systemic BRAF inhibition, indicating that the tumor rapidly becomes less sensitive to the drug during early treatment.

Concurrently, there is a strong rise in efficacy of the EGFR inhibitor erlotinib and PDGFR inhibitor imatinib, as indicated by a near doubling of the apoptotic response. Palbociclib (Cdk4/6 inhibitor) and BYL719 (PI3K inhibitor) also induce significantly increased apoptotic responses in the treated tumors. The efficacies of the other 7 targeted inhibitors tested do not change appreciably during the first 4 days of systemic BRAF inhibition (Figs. 2 and 3).

Adaptive responses during later stages of prolonged BRAF inhibition

We observe a further decrease in sensitivity to PLX4720 following 15 days of continuous BRAF inhibition. PLX4720 is mostly ineffective at this time point in inducing apoptosis (AI = 8%). Tumor sensitivity to doxorubicin appears to be largely unaffected by BRAF inhibition, showing only a slight decrease at high apoptosis levels compared with before BRAF treatment (AI = 49% vs. 52%; Fig. 3).

In contrast, tumors become significantly more sensitive to vorinostat and Byl719 on day 16, with both increasing in apoptosis induction by more than 50% compared with treatment at day 4. Erlotinib and imatinib remain at high levels of potency, but the remaining targeted agents stagnate or exhibit decreasing potency, such as palbociclib, crizotinib, and PD173074 (Fig. 3).

Identification of the most prominent bypass mechanisms

Systemic inhibition of BRAF by PLX4720 is observed to drastically increase susceptibility to several other targeted inhibitors. The most prominent increases occur in erlotinib targeting EGFR and imatinib targeting PDGFR, BYL719 targeting PI3K, and vorinostat targeting HDAC. Each of these compounds becomes 50% to 100% more efficacious in inducing intratumor apoptosis during systemic BRAF inhibition (Fig. 3B).

The tumor's sensitivity to local treatment with PLX4720 decreases greatly during systemic BRAF inhibition from AI = 40% prior to treatment to AI = 9% at 16 days. This is likely explained by the systemic BRAF inhibition rendering the remaining tumor cells increasingly resistant to this drug. Interestingly, however, the cytotoxic DNA-damaging agent doxorubicin exhibits similarly high apoptosis induction both before and after systemic BRAF inhibition. Cytotoxic DNA-damaging chemotherapy, such as doxorubicin, may be more agnostic to adaptive in signaling pathways induced by BRAF inhibition.

Investigation into the mechanisms regulating drug sensitivity after long-term BRAF inhibition

In vitro tests were performed to better understand the mechanistic basis for the efficacy enhancements observed in vivo. First, we performed survival assays on A375 cells treated for 7 days with increasing doses of PLX4720 and then challenged with an additional drug for 24 hours, focusing on four drugs, which showed the highest efficacy enhancement effects in vivo: imatinib, Byl719, erlotinib, and vorinostat. Treatment with all four drugs subsequent to PLX 4720 pretreatment led to decreases in cell survival that were significant relative to treatment with vehicle, or PLX4720 alone (Fig. 4A–D). The highest degree of efficacy enhancement is observed for vorinostat and Byl719 (Fig. 4A and B). Imatinib and erlotinib induced comparatively lower overall cell death but nonetheless exhibited an increased effect with PLX4720 (Fig. 4C and D). In total, PLX4720 treatment sensitized cells in vitro to each of the four therapies identified in vivo as exhibiting the greatest increase in potency, leading to enhanced inhibitory effects on cell growth.

Figure 4.

In vitro validation and investigation into mechanisms driving enhanced combined drug responses: survival curve for A375 cells treated with multiple concentrations of PLX4270 (0.1 or 1 μmol/L) for 7 days and then treated with vorinostat (A), Byl719 (B), imatinib (C), or erlotinib (D) for 24 hours. Representative Western blots and quantifications for pERK Y204 (E and F), pAkt473 (G and H), and pRB S807/811 (I and J) using the same drug treatment paradigm. Data were pooled from three separate experiments. Results show mean ± SEM, significance determined by one way ANOVA. *, P < 0.5; **, P < 0.01; ***, P < 0.005.

Figure 4.

In vitro validation and investigation into mechanisms driving enhanced combined drug responses: survival curve for A375 cells treated with multiple concentrations of PLX4270 (0.1 or 1 μmol/L) for 7 days and then treated with vorinostat (A), Byl719 (B), imatinib (C), or erlotinib (D) for 24 hours. Representative Western blots and quantifications for pERK Y204 (E and F), pAkt473 (G and H), and pRB S807/811 (I and J) using the same drug treatment paradigm. Data were pooled from three separate experiments. Results show mean ± SEM, significance determined by one way ANOVA. *, P < 0.5; **, P < 0.01; ***, P < 0.005.

Close modal

We next investigated the mechanism by which these drugs may combine for greater efficacy with PLX4720, focusing on three signaling pathways known to be involved in driving tumor cell proliferation in BRAF-mutant cancers: MAPK (via ERK phosphorylation), Akt (e.g., by PI3K activation), and Rb. Following long-term PLX4720 treatment, imatinib leads to a further decrease in pERK (Fig. 4E and F), an effect not observed for the other agents. Analogously, AKT signaling is further decreased significantly by PI3K inhibition via BYL719 (Fig. 4G and H). Finally, after PLX4720, the HDAC inhibitor vorinostat led to a significant decrease in RB phosphorylation relative to each treatment alone (Fig. 4I and J).

It is now well established that resistance to targeted therapies can emerge through the activation of secondary bypass signaling pathways, which can be further targeted by the use of second-line combination therapies. Given the complex nature and heterogeneity of resistance mechanisms, such studies are generally carried out in vitro, with single combinations carried out in vivo. Here, for the first time, we demonstrate the ability to study multiple drug combinations in vivo simultaneously to measure how tumors adapt functionally during systemic standard-of-care treatment and to rapidly assess the potential for more effective combinations in inducing tumor cell death. The examination of adaptive phenotypic drug responses using microdevices allows testing of several molecular hypotheses directly for each individual tumor, which broadens its potential clinical applicability as a tool for personalized medicine.

This study revealed that long-term PLX4720 treatment sensitized tumors specifically to BYL719, erlotinib, vorinostat, and imatinib. Further in vitro studies confirmed these results, and the signaling pathways driving the enhanced efficacy were investigated. Reactivation of the ERK pathway after Braf inhibitor treatment is a well-established resistance pathway (12). We find decreased ERK phosphorylation in cells treated with PLX4720 and imatinib, suggesting this combination may be effective in bypassing resistance to Braf inhibition. Imatinib has been shown to cause decreased pERK and pAkt in combination with BRAF inhibitors due to suspected adaptive responses in PDGFR signaling, leading to imatinib now being tested clinically in melanoma (13). Similarly, PI3K activation has also previously been suspected as a major potential bypass mechanism in BRAF-treated tumors (11). We show that pretreatment with PLX4270 can sensitize cells to BYL719, leading to a significant decrease in pAkt, relative to BYL719 treatment alone. Recent studies have shown that resistance to MAPK inhibition can arise through decreased shedding of several RTKs, including AXL, leading to increased activity via Akt and c-Jun (14). Given the enhanced effect of Braf and PI3K/Akt inhibitors (15), clinical trials with combinations of BRAF and PI3K inhibition are currently underway for metastatic melanoma. Finally, we show that pretreatment with PLX4720 can sensitize melanoma cells to the HDAC inhibitor vorinostat via decreased phosphorylation of Rb. HDAC inhibitors have been shown to promote cell death by activation of the proapoptotic protein Bim and the oncogenic Rb–E2F1 pathway (16). Upregulation of MITF has been implicated as a consequence of BRAF inhibitor treatment in some tumors with resistance being driven by BCL2A1 expression and a switch to oxidative phosphorylation (17–19). HDAC inhibition has been suggested as a strategy for decreasing MITF expression, and could further help explain the increased combined effect of these two drugs (20). Among other agents tested, YH239-EE and 17-DMAG each exhibit a moderate increase in sensitivity, but the levels of apoptosis remain below 30% at all treatment time points. Several drug compounds exhibit low and stable intratumor apoptotic responses that do not vary significantly throughout systemic BRAF inhibition. These compounds include crizotinib, PD173074, palbociclib, and lapatinib. SCH772984 has relatively high effect (AI = 25%), but it remains stagnant throughout treatment.

It is noteworthy that adaptive sensitizations to targeted therapies emerge at different time points during systemic treatment. For example, all of the increased sensitivity to erlotinib is observed within 4 days of treatment start, with no additional increase in the remaining 12 days. For imatinib and Byl719, greater than 50% of the total increase in drug efficacy also occurs within the first 4 treatment days. Sensitivity to vorinostat does not, however, change significantly during the initial 4 days of treatment. Here, approximately 90% of the observed sensitization occurs during the later stages of systemic BRAF therapy. Such information could be used clinically not only to choose which drugs should comprise a combination regimen, but also at which time points they should be combined for optimal efficacy.

To our knowledge, this is the first in vivo study that examines phenotypic drug response to a large number of targeted agents over the course of systemic treatment and is capable of observing parallel drug phenotypes during tumor evolution. Rather than waiting for resistance mechanisms to manifest at the time of disease progression, when the primary targeted therapy is no longer inhibiting the primary targeted pathway, this approach enables deployment of combination therapy at a time when synergistic interactions can be leveraged.

The presented technology enables a novel understanding of emerging tumor drug resistance by mapping how a given tumor responds to a range of other agents while the most prominent and actionable aberration is targeted. Clinically, this assay could be used in patients to discover which drugs should be added at a given time point to the standard-of-care treatment to achieve optimal antitumor effect of the combination therapy.

No potential conflicts of interest were disclosed.

Conception and design: O. Jonas, M.J. Oudin, M.J. Cima, K.T. Flaherty, R. Langer

Development of methodology: O. Jonas, T. Kosciuk, M.J. Cima

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): O. Jonas, M.J. Oudin, T. Kosciuk, M. Whitman

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): O. Jonas, M.J. Oudin, M. Whitman, K.T. Flaherty

Writing, review, and/or revision of the manuscript: O. Jonas, M.J. Oudin, T. Kosciuk, M.J. Cima, K.T. Flaherty, R. Langer

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M. Whitman

Study supervision: F.B. Gertler, M.J. Cima, R. Langer

This work was supported by the Koch Institute Frontier Research Program through the Kathy and Curt Marble Cancer Research Fund and the Koch Insitute Frontier Research Fund, Ludwig Center for Molecular Oncology, Prostate Cancer Foundation, Koch Institute NCI core grant P30-CA14051 and by NCI grant R21-CA177391.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Chapman
PB
,
Hauschild
A
,
Robert
C
,
Haanen
JB
,
Ascierto
P
,
Larkin
J
, et al
Improved survival with vemurafenib in melanoma with BRAF V600E mutation
.
N Engl J Med
2011
;
364
:
2507
16
.
2.
Jang
S
,
Atkins
MB
. 
Treatment of BRAF-mutant melanoma: the role of vemurafenib and other therapies
.
Clin Pharmacol Ther
2014
;
95
:
24
31
.
3.
Lito
P
,
Rosen
N
,
Solit
DB
. 
Tumor adaptation and resistance to RAF inhibitors
.
Nat Med
2013
;
19
:
1401
9
.
4.
Poulikakos
PI
,
Rosen
N
. 
Mutant BRAF melanomas–dependence and resistance
.
Cancer Cell
2011
;
19
:
11
5
.
5.
Sun
C
,
Wang
L
,
Huang
S
,
Heynen
GJJE
,
Prahallad
A
,
Robert
C
, et al
Reversible and adaptive resistance to BRAF(V600E) inhibition in melanoma
.
Nature
2014
;
508
:
118
22
.
6.
Straussman
R
,
Morikawa
T
,
Shee
K
,
Barzily-Rokni
M
,
Qian
ZR
,
Du
J
, et al
Tumour micro-environment elicits innate resistance to RAF inhibitors through HGF secretion
.
Nature
2012
;
487
:
500
4
.
7.
Wilson
TR
,
Fridlyand
J
,
Yan
Y
,
Penuel
E
,
Burton
L
,
Chan
E
, et al
Widespread potential for growth-factor-driven resistance to anticancer kinase inhibitors
.
Nature
2012
;
487
:
505
9
.
8.
Nakasone
ES
,
Askautrud
HA
,
Kees
T
,
Park
J-H
,
Plaks
V
,
Ewald
AJ
, et al
Imaging tumor-stroma interactions during chemotherapy reveals contributions of the microenvironment to resistance
.
Cancer Cell
2012
;
21
:
488
503
.
9.
Jonas
O
,
Landry
HM
,
Fuller
JE
,
Santini
JT
 Jr
,
Baselga
J
,
Tepper
RI
, et al
An implantable microdevice to perform high-throughput in vivo drug sensitivity testing in tumors
.
Sci Transl Med
2015
;
7
:
1
12
.
10.
Barretina
J
,
Caponigro
G
,
Stransky
N
,
Venkatesan
K
,
Margolin
AA
,
Kim
S
, et al
The Cancer Cell Line Encyclopedia enables predictive modelling of anticancer drug sensitivity
.
Nature
2012
;
483
:
603
7
.
11.
Greger
JG
,
Eastman
SD
,
Zhang
V
,
Bleam
MR
,
Hughes
AM
,
Smitheman
KN
, et al
Combinations of BRAF, MEK, and PI3K/mTOR inhibitors overcome acquired resistance to the BRAF inhibitor GSK2118436 dabrafenib, mediated by NRAS or MEK mutations
.
Mol Cancer Ther
2012
;
11
:
909
20
.
12.
Van Allen
EM
,
Wagle
N
,
Sucker
A
,
Treacy
DJ
,
Johannessen
CM
,
Goetz
EM
, et al
The genetic landscape of clinical resistance to RAF inhibition in metastatic melanoma
.
Cancer Discov
2014
;
4
:
94
109
.
13.
Flaherty
KT
,
Hamilton
BK
,
Rosen
MA
,
Amaravadi
RK
,
Schuchter
LM
,
Gallagher
M
, et al
Phase I/II trial of imatinib and bevacizumab in patients with advanced melanoma and other advanced cancers
.
Oncologist
2015
;
20
:
1
8
.
14.
Miller
MA
,
Oudin
MJ
,
Sullivan
RJ
,
Wang
SJ
,
Meyer
AS
,
Im
H
, et al
Reduced proteolytic shedding of receptor tyrosine kinases is a post-translational mechanism of kinase inhibitor resistance
.
Cancer Discov
2016
;
6
:
382
99
.
15.
Sweetlove
M
,
Wrightson
E
,
Kolekar
S
,
Rewcastle
GW
,
Baguley
BC
,
Shepherd
PR
, et al
Inhibitors of pan-PI3K signaling synergize with BRAF or MEK inhibitors to prevent BRAF-mutant melanoma cell growth
.
Front Oncol
2015
;
5
:
135
.
16.
Zhao
Y
,
Tan
J
,
Zhuang
L
,
Jiang
X
,
Liu
ET
,
Yu
Q
. 
Inhibitors of histone deacetylases target the Rb-E2F1 pathway for apoptosis induction through activation of proapoptotic protein Bim
.
Proc Natl Acad Sci U S A
2005
;
102
:
16090
5
.
17.
Haq
R
,
Yokoyama
S
,
Hawryluk
EB
,
Jönsson
GB
,
Frederick
DT
,
McHenry
K
, et al
BCL2A1 is a lineage-specific antiapoptotic melanoma oncogene that confers resistance to BRAF inhibition
.
Proc Natl Acad Sci U S A
2013
;
110
:
4321
6
.
18.
Haq
R
,
Shoag
J
,
Andreu-Perez
P
,
Yokoyama
S
,
Edelman
H
,
Rowe
GC
, et al
Oncogenic BRAF regulates oxidative metabolism via PGC1α and MITF
.
Cancer Cell
2013
;
23
:
302
15
.
19.
Haq
R
,
Fisher
DE
,
Widlund
HR
. 
Molecular pathways: BRAF induces bioenergetic adaptation by attenuating oxidative phosphorylation
.
Clin Cancer Res
2014
;
20
:
2257
63
.
20.
Konieczkowski
DJ
,
Johannessen
CM
,
Abudayyeh
O
,
Kim
JW
,
Cooper
ZA
,
Piris
A
, et al
A melanoma cell state distinction influences sensitivity to MAPK pathway inhibitors
.
Cancer Discov
2014
;
4
:
816
27
.