Abstract
Tyrosine kinase inhibitors are effective treatments for cancers. Knowing the specific kinase mutants that drive the underlying cancers predict therapeutic response to these inhibitors. Thus, the current protocol for personalized cancer therapy involves genotyping tumors in search of various driver mutations and subsequently individualizing the tyrosine kinase inhibitor to the patients whose tumors express the corresponding driver mutant. While this approach works when known driver mutations are found, its limitation is the dependence on driver mutations as predictors for response. To complement the genotype approach, we hypothesize that a phosphoarray platform is equally capable of personalizing kinase inhibitor therapy. We selected head and neck squamous cell carcinoma as the cancer model to test our hypothesis. Using the receptor tyrosine kinase phosphoarray, we identified the phosphorylation profiles of 49 different tyrosine kinase receptors in five different head and neck cancer cell lines. Based on these results, we tested the cell line response to the corresponding kinase inhibitor therapy. We found that this phosphoarray accurately informed the kinase inhibitor response profile of the cell lines. Next, we determined the phosphorylation profiles of 39 head and neck cancer patient derived xenografts. We found that absent phosphorylated EGFR signal predicted primary resistance to cetuximab treatment in the xenografts without phosphorylated ErbB2. Meanwhile, absent ErbB2 signaling in the xenografts with phosphorylated EGFR is associated with a higher likelihood of response to cetuximab. In summary, the phosphoarray technology has the potential to become a new diagnostic platform for personalized cancer therapy.
Keywords: phosphoarray, head and neck squamous cell carcinoma, kinase inhibitors, personalized medicine, cetuximab response
Introduction
Imatinib is the first tyrosine kinase inhibitor (TKI) that directly targeted an oncogenic driver mutant. This drug showed unprecedented success in the treatment of chronic myleogenous leukemia (1). Since then, many kinase inhibitors targeting different oncogenic kinases were developed. A few of these drugs showed equally impressive efficacy, for instance, crizotinib for the non small cell lung cancers (NSCLC) that harbored the EML4-ALK translocation (2), vemurafenib for the BRAF V600E mutated melanoma (3), erlotinib for the NSCLC that harbored activating EGFR kinase mutations (4) or vandetanib for the hereditary medullary thyroid cancer with underlying RET mutation (5). Like imatinib, the common theme around these success stories is that the TKIs specifically targeted the oncogenic mutants that drive the underlying cancers. Thus, recent effort has been focused on profiling the genetic landscape of tumors to identify potential druggable targets, thereby increasing the efficacy of kinase inhibitor therapies.
With the rapid advance in sequencing technologies, high throughput screening of mutation drivers by next generation sequencing (NGS) is now a commercially available service for personalized cancer therapy. There are many anecdotal cases that utilized the NGS platform to identify driver mutations in cancer patients for novel TKI therapy (6–10). In some cases, the diagnostic was successful in personalizing the right TKIs for the right patients. For instance, when a 41 year old woman with refractory, progressive sarcoma ran out of therapeutic options, NGS identified a novel TRK receptor fusion product, LMNA-NTRK1, in her original tumor. She was subsequently enrolled in a phase I trial of a new pan-TRK inhibitor, LOXO-101. After five cycles of LOXO-101, there was complete resolution of her metastatic diseases (11). Similarly, after MET exon 14 mutations were identified in 0.6% of lung adenocarcinoma by NGS, three patients with tumors harboring these MET mutants were treated with MET directed therapies via clinical trials. All three demonstrated partial responses (12).
Despite these success stories, the NGS platform has limitations as a personalized diagnostic. First, it might reveal many passenger mutations that are not drivers of the tumor. Second, bearing driver mutants does not necessarily translate into response to the corresponding TKIs. For instance, vemurafenib did not produce a dramatic response in the treatment of BRAF V600E mutated colorectal cancer (13). Third, low mutation rates in some cancers like pediatric tumors (14) might limit the usefulness of NGS as a personalized diagnostic. Fourth, there might not be mutation drivers of a known target in the tumor. For example, EGFR is a known target for head and neck squamous cell carcinomas (HNSCC), but HNSCC rarely carried activating EGFR kinase mutations (15, 16). Finally, there might be other mechanisms of altering oncoprotein function/activity that NGS diagnostic is not able to identify. Such mechanisms might include overexpression, impaired degradation, defective negative feedback loop or constitutive activation. To improve the genotype approach, we hypothesized that a diagnostic that examine the activity of multiple kinases simultaneously might complement the NGS platform for better selection of the right patient for the right TKI.
Phosphoarray is a high throughput screening tool that examines the activities of multiple kinases simultaneously. One commonly used array is called the human phospho-receptor tyrosine kinase (RTK) array. Capture and control antibodies were spotted in duplicate on nitrocellulose membranes. When cell/tumor lysates were incubated with this array, both phosphorylated and unphosphorylated RTKs would bind. The active receptors would then be detected by chemiluminescence using a horseradish peroxidase conjugated anti-phospho-tyrosine antibody. The phospho-RTK array can examine the phosphorylation status of 49 RTKs in the lysates simultaneously. While this array has been used in the laboratory setting to identify molecular pathway changes (17–21), it has not been tested as a diagnostic to predict tumor response to TKI therapy. However, when it was used retrospectively to identify pathway changes in primary tumors, the array results seem to correlate with patient’s response to the TKI, sunitinib (22, 23). In the four refractory thymic carcinoma patients who demonstrated response to sunitinib, the phospho-RTK array identified KIT, a target of sunitinib, as active in their tumors (22). Similarly, the two patients with progressive metastatic alveolar soft part sarcoma who showed partial responses to sunitinib had active PDGFR on the array (23). These findings implied that the phospho-RTK array might be useful as a diagnostic to predict individual tumor response to TKI therapy. In this report, we demonstrated that the phospho-RTK array can inform the TKI response profile of head and neck cancer cell line and patient derived xenograft (PDX) model.
Materials and Methods
Cell lines, reagents and antibodies
The HNSCC cell lines (SCC9, SCC15, CAL27, SCC25 and MDA1386) were obtained, characterized, grown in media and condition as previously described (24). All of the cell lines have been authenticated by short tandem repeat profiling within six months of passage. The phospho-receptor tyrosine kinase array was purchased (ARY001B, R&D Systems, Minneapolis, MN). The array layout of the 49 RTKs were shown in the Supplemental Figure (Fig. S5). The following small molecular TKIs were purchased (Selleck Chemicals): (1) JNJ-38877605, a highly selective, ATP-competitive inhibitor of c-MET (25); (2) NVP-AEW541, a potent inhibitor of IGF-1R with IC50 of 86 nM (26); (3) OSI-744/erlotinib HCl, a FDA approved EGFR inhibitor and (4) STI-571/imatinib, a multi-target inhibitor of v-Abl, c-Kit and PDGFR. The TKIs were reconstituted in DMSO solvent as per manufacture recommendation.
Collection and processing of primary HNSCC tumors
Snap frozen primary HNSCC were collected through the Cooperative Human Tissue Network and the HNSCC tumor lysates were prepared for biochemical analyses by the homogenization method as previously described (15, 24). The snap frozen primary HNSCC were accrued as de-identified samples with no link to clinical information. On the other hand, the human HNSCC that used to establish the PDX were clinically annotated. All patients included in this study had given written informed consent. The collection of patients’ materials for the PDX experiments and for the biochemical analyses was approved by the local Institutional Review Board of Charité University Medicine, Germany (EA4/019/12) and of the Stony Brook University respectively.
XTT proliferation assay
XTT proliferation assays were performed as previously described (24). Briefly, cells were seeded at 104 cells/well in a 96 well plate in quintuplicate. The cells were treated the next day with increasing concentration of the corresponding tyrosine kinase inhibitor (1 – 10 uM) or DMSO control. Activated-XTT reagent was prepared and added to the cells the following day as per protocol. Cell proliferation rates were determined as previously described (24). The proliferation rate of untreated cells served as baseline for comparison to that of treated cells. The percent of cell growth inhibition equaled one minus the proliferation rate of treated cells divided by that of untreated cells. A minimum of three independent experiments were performed at each concentration of treatment.
PDX treatment study and correlation analyses
Fresh tumor materials from patients who consented to the PDX treatment study was subcutaneously transplanted into NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ (NSG) mice as previously described (27). Groups of 5–6 animals were randomized to treatment with cetuximab or saline as control according to schedule as described (27). Tumor measurement was done at two dimensions with a sliding caliper twice a week during the three-week period of treatment. Treatment was initiated at a tumor size of 100 – 150 mm3. Therefore, the experiments were performed as regression studies resembling the clinical situation. Individual tumor volumes (V) were calculated by the formula: V = ([width]2 * length)/2. Change in tumor volume during the course of treatment was defined as the expression value most comparable to clinical tumor evaluation. Treatment response was defined by the relative tumor volume (RTV) which equaled tumor volume at the end of cetuximab treatment divided by that at the beginning of treatment. RTV of 0 is complete response (CR); RTV below 0.8 is partial response (PR); RTV between 0.8–1.2 is stable disease (SD) and RTV above 1.2 is progressive disease (PD). For the progressive tumor, the growth curve was examined and compared to the saline control. Primary resistance is defined as tumor growth curve overlapping the control curve, while secondary resistance is defined as increased tumor growth velocity at a later time point after an initial PR or SD. The different types of treatment response were illustrated in Supplemental Figure 1 (Fig. S1). The investigators (E.L.C. and J.K.) who performed the phosphoarray were blinded to the cetuximab treatment response until the array results were analyzed. All animal experiments were carried out in accordance with the United Kingdom coordinating committee on cancer research regulations for the welfare of animals and the German Animal Protection Law, and the protocols were also approved by the local responsible authorities (LaGeSoBerlin, A0452/08).
Biochemical analysis
Cell, PDX and primary tumor lysates were analyzed for receptor tyrosine kinase signaling using the phospho-RTK arrays as per manufacture recommendation (R&D Systems). The following guidelines were established for the interpretation of the array result: (1) Positive hit of a target (+) is defined as signal intensity stronger than or equally as strong as the positive controls on the same array blot; (2) Signals that are visibly about half the intensity of the positive controls on the same array blot will be recorded as intermediate (+/−); and (3) Signals that are less than half the intensity of the positive controls or nonvisible will be classified as negative (−). An illustration of the phosphoarray interpretation was shown in the Supplemental Figures (Figs. S2 and S4).
Statistics Analysis
Statistical analyses were performed using SPSS Statistics 16.0 (SPSS Inc., Chicago, IL). Comparison between the different drug treatment groups was performed using ANOVA. Fisher exact test was used to examine the association between pEGFR/pErbB2 status and cetuximab treatment response. Level of statistical significance is 5%.
Results
RTK phosphoarray predicts head and neck cancer cell response to kinase inhibitors
Using the phospho-RTK array, we determined that the HNSCC cell line, MDA1386, had activated EGFR and MET (Fig. 1A). We also noticed intermediate phosphorylation signals from IGF-1R and Axl, but no PDGFR activity in this cell line (Fig. 1A). Based on the array result, we treated the MDA1386 cells with the corresponding TKI. As expected, imatinib had no inhibitory effect on MDA1386 at all concentrations (Fig. S3). On the other hand, MET TKI treatment resulted in significant cell growth inhibition at 5 and 10 uM (Fig. 1B). While IGF-1R TKI also had a significant inhibitory effect on cell growth at 10 uM, it had no effect at 5 uM (Fig. 1C). This was predicted given the weak IGF-1R phosphorylation signal detected (Fig. 1A). Even though EGFR was highly phosphorylated (Fig. 1A), erlotinib had limited inhibitory effect on MDA1386 cell growth (Fig. 1D). This was also predicted because MET signaling is known to mediate EGFR TKI resistance (28–32). To determine if dual EGFR and MET inhibition resulted in a greater therapeutic effect than MET TKI alone, we treated the MDA1386 cells with increasing concentrations of both inhibitors. As shown, the therapeutic effect of dual inhibition was significantly greater than that of either inhibitor alone (Fig. 1D). This implied a synergism between the two TKIs. In summary, the phospho-RTK array accurately informed the TKI response profile of the MDA1386 cell line.
Fig. 1. MDA1386 TKI response profile.
(A) Phospho-RTK array analysis of MDA1386 cell lysate. The positive (+) controls are the built–in reference spots at the three corners of the array blot. Black arrows pointed to the positions corresponding to the respective RTK on the blot. Noted the strong EGFR and MET phosphorylation signals, the weaker IGF-1R and Axl signals and the absent PDGFR signal. (B) MDA1386 cell response to the MET TKI at three different concentrations (1, 5 and 10 uM, n = 3 at each concentration for each treatment). Tx: treatment; NS: not significant. Error bars represent ±2 standard errors. (C) MDA1386 cell response to the IGF-1R TKI at three different concentrations (1, 5 and 10 uM, n = 3 at each concentration for each treatment). Tx: treatment; NS: not significant. Error bars represent ±2 standard errors. (D) MDA1386 cell response to the dual inhibition of MET and EGFR in comparison to MET or EGFR inhibition alone at three different concentrations (1, 5 and 10 uM, n = 3 at each concentration for each treatment). Noted the increase in cell growth inhibition with dual MET and EGFR TKI at 10 uM. Tx: treatment; NS: not significant. Error bars represent ±2 standard errors.
Next, we determined the RTK phosphorylation profiles of four additional HNSCC cell lines. Surprisingly, all of the cell lines had similar profiles (Fig. 2A and S2). Since SCC25 had a weaker MET phosphorylation signal than MDA1386 (Figs. 1A and 2A), we decided to compare SCC25 response to MET TKI with that of MDA1386. While MET TKI inhibited the cell proliferation of both cell lines at 10 uM, its effect on MDA1386 was significantly stronger than that on SCC25 (Fig. 2B). We also tested SCC25 response to EGFR TKI. Based on the array result, we speculated that SCC25 would be more responsive to EGFR TKI than MDA1386 because MET activation was weaker in this cell line. Indeed, erlotinib resulted in a significantly greater growth inhibition in SCC25 than MDA1386 (Fig. 2C). On the other hand, we anticipated a greater effect of IGF-1R TKI on SCC25 than MDA1386 as SCC25 had a stronger IGF-1R phosphorylation signal. As anticipated, IGF-1R TKI resulted in a higher percentage of growth inhibition in SCC25 than MDA1386 (Fig. 2D). In summary, the phospho-RTK array informed the differential response of two HNSCC cell lines to kinase inhibitor therapies.
Fig. 2. SCC25 cell line response to TKI.
(A) Phospho-RTK array analysis of SCC25 cell lysate. Noted the strong EGFR and IGF-1R phosphorylation signals and the weaker MET signal. (B) Comparison of the degree of cell growth inhibition from baseline by MET TKI at 10 uM between the two cell lines (MDA1386 and SCC25, n = 3). Error bars represent ±2 standard errors. (C) Comparison of the degree of cell growth inhibition from baseline by EGFR TKI at 10 uM between MDA1386 and SCC25, n = 3. Error bars represent ±2 standard errors. (D) Comparison of the degree of cell growth inhibition from baseline by IGF-1R TKI at 5 uM between MDA1386 and SCC25, n = 3. Error bars represent ±2 standard errors.
RTK phosphoarray predicts head and neck cancer PDX response to cetuximab
Using the phospho-RTK array, we determined the RTK phosphorylation profiles of 39 treatment naïve PDX tumors (Table I). In contrast to the cell line data, none of the PDX had strong MET phosphorylation signal. While phosphorylated EGFR was the predominant signal in the majority (89.7%, 35/39) of PDX, 28.2% (11/39) had ErbB2 signaling (Fig. 3A and Table I). We suspected that ErbB2 activation in the PDX is the result of EGFR-ErbB2 heterodimer signaling because (1) ErbB2 amplification rarely occurred in HNSCC (33) and no ErbB2 mutations were detected in any of the PDX (27); (2), all of the pErbB2+ PDX (n = 11) had phosphorylated EGFR, but none of the pEGFR– PDX (n = 4) had pErbB2 (Table I); (3) ErbB2 receptor has no known ligand for its activation (34). Based on these evidences, the only plausible explanation for ErbB2 activation in the PDX is EGFR-ErbB2 heterodimer signaling. Next, we correlated pEGFR signal with response to cetuximab treatment. Interestingly, there was a highly significant association between primary resistance to cetuximab and negative pEGFR signal in the pErbB2− PDX (Fig. 3B). The positive and negative predictive values of pEGFR status in predicting response in this PDX cohort were 83.3 and 100% respectively. Then, we examined the association between pErbB2 signal and response to cetuximab in the pEGFR+ PDX. While the positive predictive value of a negative pErbB2 result in predicting cetuximab response in this PDX cohort was 83.3%, a positive pErbB2 signal did not necessarily predict cetuximab resistance (Fig. 3C). The association between pEGFR/pErbB2 signal and the clinical/pathological feature of each PDX was examined (Table II). pErbB2 signal or the lack of pEGFR signal does not correlate to stage, tumor grade or location. Taken together, the phospho-RTK array informed the cetuximab treatment response in the HNSCC PDX.
Table I.
Summary of HNSCC PDX RTK phosphorylation profile and response to cetuximab.
| HNSCC PDX | pEGFR | pErbB2 | Cetuximab Response |
|---|---|---|---|
| 11143 | + | + | PR |
| 10883 | + | − | PR |
| 11204A | + | − | SD |
| 11857B | + | − | CR |
| 11841 | + | − | 20 resistance |
| 11527A | − | − | 10 resistance |
| 12346 | + | − | SD |
| 11873 | + | − | SD |
| 9876 | + | + | 10 resistance |
| 9897 | + | + | 10 resistance |
| 10114 | + | + | CR |
| 10309 | + | + | 10 resistance |
| 10321 | + | + | 10 resistance |
| 10621 | + | + | 10 resistance |
| 10913 | + | +/− | SD |
| 10924 | +/− | − | 10 resistance |
| 11303 | + | − | 10 resistance |
| 11452 | +/− | − | 10 resistance |
| 11857A | +/− | − | 10 resistance |
| 13194 | + | + | SD |
| 11178 | + | − | 10 resistance |
| 11366 | + | − | PR |
| 10645 | + | − | SD |
| 11646 | + | − | PR |
| 12048 | + | − | PR |
| 11057 | + | − | SD |
| 11554 | + | − | CR |
| 11865 | + | − | PR |
| 11365 | + | − | SD |
| 11553 | + | − | SD |
| 11896 | + | + | PR |
| 11857 | + | − | 10 resistance |
| 11931 | + | − | SD |
| 11498 | + | − | PR |
| 11318 | + | + | 10 resistance |
| 10647 | + | − | 10 resistance |
| 10890 | + | +/− | 20 resistance |
| 11555 | + | − | PR |
| 11647 | + | + | PR |
Fig. 3. RTK phosphorylation profile of HNSCC PDX.
(A) Representative phospho-RTK array blots of four HNSCC PDX. Noted the strong EGFR and ErbB2 phosphorylation signals in PDX 11143 and the absent EGFR phosphorylation signal in PDX 11527A. (B) The association between pEGFR and cetuximab treatment response in the pErbB2 negative HNSCC PDX (n = 28). Cetuximab response included CR + PR + SD + the initially responsive tumor that later developed resistance (i.e. secondary resistance). (C) The association between pErbB2 and cetuximab treatment response in the pEGFR+ HNSCC PDX (n = 35). Cetuximab response included CR + PR + SD + the initially responsive tumor that later developed resistance (i.e. secondary resistance).
Table II.
Summary of HNSCC PDX clinical and pathological features.
| PDX ID | TNM | stage | grading | age | site of tumor origin |
gender |
|---|---|---|---|---|---|---|
| 9876 | T3N2cM0 | IVA | G3 | 62 | hypopharynx | male |
| 9897 | T2N2bM0 | IVA | G3 | 58 | hypopharynx | male |
| 10114 | T3N0M0 | III | G3 | 52 | oral cavity | male |
| 10309 | T4N2cM0 | IVA | G3 | 55 | oropharynx | male |
| 10321 | T2N0M0 | II | G2 | 65 | oral cavity | male |
| 10621 | T2N2bM0 | IVA | G3 | 61 | oropharynx | male |
| 10645 | T2N2cM0 | IVA | G2 | 69 | oral cavity | male |
| 10647 | T2N0M0 | II | G2 | 65 | oral cavity | male |
| 10883 | T4N0M0 | IVA | G2 | 52 | oropharynx | male |
| 10890 | T2N0M0 | II | NA | NA | oropharynx | female |
| 10913 | T4N2bM0 | IVA | G2 | 50 | oral cavity | male |
| 10924 | T3N2CM0 | IVA | G2 | 65 | hypopharynx | male |
| 11057 | T1N0M0 | I | G2 | 57 | oral cavity | male |
| 11143 | T2N2bM0 | IVA | NA | 82 | oropharynx | male |
| 11178 | T2N1M0 | III | G3 | 60 | oral cavity | male |
| 11303 | T3N1M0 | IVA | G2 | 75 | oropharynx | male |
| 11318 | T2N2bM0 | IVA | G3 | 61 | oropharynx | male |
| 11365 | T4bN2bM0 | IVA | G2 | 59 | oral cavity | female |
| 11366 | T2N0M0 | II | G2 | 63 | oral cavity | male |
| 11452 | T2N0M0 | II | G2 | 75 | oral cavity | male |
| 11498 | T2N2bM0 | IVA | G2 | 67 | oropharynx | male |
| 11553 | T4bN2bM0 | IVA | G2 | 59 | oral cavity | female |
| 11554 | T4N0M0 | IVA | G2 | 68 | oral cavity | female |
| 11555 | T4aN2bM0 | IVA | G2 | 75 | oral cavity | female |
| 11646 | T4aN2cM0 | IVA | G2 | 71 | oral cavity | male |
| 11647 | T4aN2cM0 | IVA | G2 | 71 | oral cavity | male |
| 11841 | T1N0M0 | I | G2 | 56 | oral cavity | female |
| 11857 | T4N2M0 | IVA | G1 | 49 | oral cavity | male |
| 11865 | T4bN2cM0 | IVA | G2 | 56 | oral cavity | male |
| 11873 | T2N2M0 | IVA | G3 | 47 | oropharynx | male |
| 11896 | T4bN2cM0 | IVA | G2 | 56 | oral cavity | male |
| 11931 | T2N2bM0 | IVA | G2 | 61 | oral cavity | male |
| 12048 | T2N2cM0 | IVA | G3 | 46 | oral cavity | male |
| 12346 | T1N2bM0 | IVA | G3 | 76 | oropharynx | male |
| 13194 | T4N2bM0 | IVA | G2 | 50 | oral cavity | male |
| 11204A | T4N2cM0 | IVA | G3 | 56 | oral cavity | female |
| 11527A | T2N2PM0 | IVA | G2 | 74 | oral cavity | male |
| 11857A | T4N2M0 | IVA | G1 | 49 | oral cavity | male |
| 11857B | T4N2M0 | IVA | G1 | 49 | oral cavity | male |
| pErbB2 + | pEGFR neg |
The RTK phosphorylation profile of primary HNSCC is different from those of HNSCC PDX and cell lines
We were surprised to see the difference in RTK signaling pattern between the HNSCC cell lines and PDX. To determine their similarity to the primary tumor, we performed phospho-RTK array analysis on nine freshly prepared primary HNSCC lysates. Unlike the cell lines, none had strong MET phosphorylation signal (Figs. 4 and S4). Like the HNSCC PDX, primary HNSCC had variable degree of EGFR phosphorylation (Figs. 4 and S4). This is consistent with our prior finding (15). Nevertheless, the percentage of primary HNSCC with weak (+/−) to undetectable (−) phosphorylated EGFR (66.7%, 6/9) was higher than that of PDX (10.3%, 4/39). In addition, none of the primary HNSCC had ErbB2 activation (Figs. 4 and S4). Not detecting phosphorylated ErbB2/MET in the primary tumors suggested that the cell line or PDX might have acquired dependence on signaling pathways outside of the primary setting. Despite this subtle signaling difference between the preclinical model and the primary tumors, the phospho-RTK array is able to distinguish them and thus might also be useful in the clinical setting.
Fig. 4. Representative phospho-RTK array blots of four primary HNSCC.
Noted that only HNSCC 59290 had strong EGFR phosphorylation signal.
Discussion
Advancing precision medicine has become a national priority. Oncology is at the forefront of this initiative. The advance of genomic technology has made it possible to sequence tumor genome in real time to inform treatment decision and bring personalized medicine to cancer patients. While NGS is a promising platform, the limitation is its dependence on driver mutations as predictors of response to novel therapy. In this report, we tested the phosphoarray as a new platform for personalizing kinase inhibitor therapy. In both in vitro and in vivo model, the phospho-RTK array was able to inform the kinase inhibitor response of HNSCC cell lines and PDX. During the course of the study, we made several interesting observations. First, the result that 66.7% of primary HNSCC did not have detectable pEGFR signal is similar to our prior finding in a larger cohort using a different detection method (60.7%, 34/56) (15). This is in sharp contrast to the much lower percentage of HNSCC PDX with undetectable pEGFR signal (10.3%). Since negative pEGFR predicted primary cetuximab resistance in the PDX model, the high percentage of pEGFR− primary tumor might explain the low cetuximab response rate in HNSCC clinical trials (35). Second, while the signaling profiles of HNSCC PDX bore close resemblance to that of the primary tumor, there were subtle differences. These differences might have been acquired during PDX passages. Thus, PDX response to targeted therapy should be carefully interpreted. Third, we noticed signaling and treatment response differences between PDX from different disease sites of the same patient (i.e. 11857A: primary tumor vs. 11857B: metastatic site) and PDX from disease at different time point of the same patient (i.e. 13194: primary tumor vs. 10913: recurrent tumor) (Table I). This finding supports assessing the molecular profile of not only tumors at different time points, but also tumors at different sites. Taken together, this study supports further development of the phosphoarray platform as a personalized diagnostic. Despite these promising results, there are several limitations with this platform. First, the TKI might not be potent enough to shut down the signaling of the target even when the right target was identified by the array. This could be secondary to a mutated target. Thus, the phosphoarray platform should be used in conjunction with the NGS platform to personalize TKI therapy. Second, there might be unknown compensatory mechanism(s) that conferred TKI resistance to the target identified. The phospho-RTK array does not include signaling pathways downstream of the RTK. Third, the array results can only be interpreted subjectively and do not take into account the differences in signaling strength. To improve on the currently available array, the next step will be to design and develop a quantitative array with expanded coverage of all potential druggable targets and resistant pathways. In conclusion, the phosphoarray technology and concept might be broadly applied to all cancer types and impact the field of personalized medicine.
Supplementary Material
Impact and novelty.
Advancing precision medicine has become a national priority. Currently, next generation sequencing is the only high throughput platform that could personalize targeted therapy. In this report, we showed for the first time that a phosphoarray platform is also capable of individualizing kinase inhibitor therapy. The results provided the proof of concept that this platform can be further developed into a diagnostic suitable for use in the clinic to inform treatment decision.
Acknowledgments
Grant Supports
This work was supported in part by research grants from the National Cancer Institute (1R21CA187554) (M.J.H. & E.L.C.) and the Sunrise Fund (E.L.C.).
Frozen primary tumors were provided by the Cooperative Human Tissue Network, which is funded by the National Cancer Institute. We thank Dr. Anjaruwee S. Nimnual for reading and editing the manuscript. We also liked to acknowledge Mrs. Patti Kelly for her effort in raising the Sunrise Fund to support this research in memory and honor of her loving daughter, Lizzie Kelly.
The abbreviations used are
- TKI
tyrosine kinase inhibitor
- HNSCC
head and neck squamous cell carcinoma
- NSCLC
non small cell lung cancers
- RTK
receptor tyrosine kinase
- EGFR
epidermal growth factor receptor
- MET
hepatocyte growth factor receptor
- IGFR
insulin growth factor receptor
- PDGFR
platelet derived growth factor receptor
- ErbB2/HER2
avian erythroblastosis oncogene B/human epidermal growth factor receptor 2
- p
phosphorylated
- PDX
patient derived xenograft
- NGS
next generation sequencing
- DMSO
dimethyl sulfoxide
- CR
complete response
- PR
partial response
- SD
stable disease
- PD
progressive disease
Footnotes
Conflict of Interest Disclosure Statement
Dr. Jens Hoffmann has ownership in and is also employed by the company, Experimental Pharmacology and Oncology Berlin-Buch GmbH. The remaining authors disclose no potential conflicts of interest.
References
- 1.Druker BJ, Guilhot F, O'Brien SG, et al. Five-year follow-up of patients receiving imatinib for chronic myeloid leukemia. N Engl J Med. 2006;355:2408–17. doi: 10.1056/NEJMoa062867. [DOI] [PubMed] [Google Scholar]
- 2.Solomon BJ, Mok T, Kim DW, et al. First-line crizotinib versus chemotherapy in ALK-positive lung cancer. N Engl J Med. 371:2167–77. doi: 10.1056/NEJMoa1408440. [DOI] [PubMed] [Google Scholar]
- 3.Chapman PB, Hauschild A, Robert C, et al. Improved survival with vemurafenib in melanoma with BRAF V600E mutation. N Engl J Med. 364:2507–16. doi: 10.1056/NEJMoa1103782. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Lynch TJ, Bell DW, Sordella R, et al. Activating mutations in the epidermal growth factor receptor underlying responsiveness of non-small-cell lung cancer to gefitinib. N Engl J Med. 2004;350:2129–39. doi: 10.1056/NEJMoa040938. [DOI] [PubMed] [Google Scholar]
- 5.Fox E, Widemann BC, Chuk MK, et al. Vandetanib in Children and Adolescents with Multiple Endocrine Neoplasia Type 2B Associated Medullary Thyroid Carcinoma. Clin Cancer Res. 19:4239–48. doi: 10.1158/1078-0432.CCR-13-0071. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Gallant JN, Sheehan JH, Shaver TM, et al. EGFR Kinase Domain Duplication (EGFR-KDD) Is a Novel Oncogenic Driver in Lung Cancer That Is Clinically Responsive to Afatinib. Cancer Discov. 5:1155–63. doi: 10.1158/2159-8290.CD-15-0654. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Falchook GS, Ordonez NG, Bastida CC, et al. Effect of the RET Inhibitor Vandetanib in a Patient With RET Fusion-Positive Metastatic Non-Small-Cell Lung Cancer. J Clin Oncol. 34:e141–4. doi: 10.1200/JCO.2013.50.5016. [DOI] [PubMed] [Google Scholar]
- 8.Chalmers ZR, Ali SM, Ohgami RS, et al. Comprehensive genomic profiling identifies a novel TNKS2-PDGFRA fusion that defines a myeloid neoplasm with eosinophilia that responded dramatically to imatinib therapy. Blood Cancer J. 5:e278. doi: 10.1038/bcj.2014.95. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Carneiro BA, Elvin JA, Kamath SD, et al. FGFR3-TACC3: A novel gene fusion in cervical cancer. Gynecol Oncol Rep. 13:53–6. doi: 10.1016/j.gore.2015.06.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Subbiah V, Berry J, Roxas M, et al. Systemic and CNS activity of the RET inhibitor vandetanib combined with the mTOR inhibitor everolimus in KIF5B-RET re-arranged non-small cell lung cancer with brain metastases. Lung Cancer. 89:76–9. doi: 10.1016/j.lungcan.2015.04.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Doebele RC, Davis LE, Vaishnavi A, et al. An Oncogenic NTRK Fusion in a Patient with Soft-Tissue Sarcoma with Response to the Tropomyosin-Related Kinase Inhibitor LOXO-101. Cancer Discov. 5:1049–57. doi: 10.1158/2159-8290.CD-15-0443. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Frampton GM, Ali SM, Rosenzweig M, et al. Activation of MET via diverse exon 14 splicing alterations occurs in multiple tumor types and confers clinical sensitivity to MET inhibitors. Cancer Discov. 5:850–9. doi: 10.1158/2159-8290.CD-15-0285. [DOI] [PubMed] [Google Scholar]
- 13.Nagaraja AK, Bass AJ. Hitting the Target in BRAF-Mutant Colorectal Cancer. J Clin Oncol. 33:3990–2. doi: 10.1200/JCO.2015.63.7793. [DOI] [PubMed] [Google Scholar]
- 14.Vogelstein B, Papadopoulos N, Velculescu VE, Zhou S, Diaz LA, Jr, Kinzler KW. Cancer genome landscapes. Science. 339:1546–58. doi: 10.1126/science.1235122. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Keller J, Shroyer KR, Batajoo SK, et al. Combination of phosphorylated and truncated EGFR correlates with higher tumor and nodal stage in head and neck cancer. Cancer Invest. 28:1054–62. doi: 10.3109/07357907.2010.512602. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Van Allen EM, Lui VW, Egloff AM, et al. Genomic Correlate of Exceptional Erlotinib Response in Head and Neck Squamous Cell Carcinoma. JAMA Oncol. 1:238–44. doi: 10.1001/jamaoncol.2015.34. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Bai Y, Li J, Fang B, et al. Phosphoproteomics identifies driver tyrosine kinases in sarcoma cell lines and tumors. Cancer Res. 72:2501–11. doi: 10.1158/0008-5472.CAN-11-3015. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Duncan JS, Whittle MC, Nakamura K, et al. Dynamic reprogramming of the kinome in response to targeted MEK inhibition in triple-negative breast cancer. Cell. 149:307–21. doi: 10.1016/j.cell.2012.02.053. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Carretero J, Shimamura T, Rikova K, et al. Integrative genomic and proteomic analyses identify targets for Lkb1-deficient metastatic lung tumors. Cancer Cell. 17:547–59. doi: 10.1016/j.ccr.2010.04.026. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Dunn EF, Iida M, Myers RA, et al. Dasatinib sensitizes KRAS mutant colorectal tumors to cetuximab. Oncogene. 30:561–74. doi: 10.1038/onc.2010.430. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Keller J, Nimnual AS, Shroyer KR, et al. Ron tyrosine kinase receptor synergises with EGFR to confer adverse features in head and neck squamous cell carcinoma. Br J Cancer. 109:482–92. doi: 10.1038/bjc.2013.321. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Strobel P, Bargou R, Wolff A, et al. Sunitinib in metastatic thymic carcinomas: laboratory findings and initial clinical experience. Br J Cancer. 103:196–200. doi: 10.1038/sj.bjc.6605740. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Stacchiotti S, Tamborini E, Marrari A, et al. Response to sunitinib malate in advanced alveolar soft part sarcoma. Clin Cancer Res. 2009;15:1096–104. doi: 10.1158/1078-0432.CCR-08-2050. [DOI] [PubMed] [Google Scholar]
- 24.Keller J, Nimnual AS, Shroyer KR, et al. Ron tyrosine kinase receptor synergises with EGFR to confer adverse features in head and neck squamous cell carcinoma. Br J Cancer. 109:482–92. doi: 10.1038/bjc.2013.321. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Eder JP, Vande Woude GF, Boerner SA, LoRusso PM. Novel therapeutic inhibitors of the c-Met signaling pathway in cancer. Clin Cancer Res. 2009;15:2207–14. doi: 10.1158/1078-0432.CCR-08-1306. [DOI] [PubMed] [Google Scholar]
- 26.Garcia-Echeverria C, Pearson MA, Marti A, et al. In vivo antitumor activity of NVP-AEW541-A novel, potent, and selective inhibitor of the IGF-IR kinase. Cancer Cell. 2004;5:231–9. doi: 10.1016/s1535-6108(04)00051-0. [DOI] [PubMed] [Google Scholar]
- 27.Klinghammer K, Raguse JD, Plath T, et al. A comprehensively characterized large panel of head and neck cancer patient-derived xenografts identifies the mTOR inhibitor everolimus as potential new treatment option. Int J Cancer. 136:2940–8. doi: 10.1002/ijc.29344. [DOI] [PubMed] [Google Scholar]
- 28.Chen CT, Kim H, Liska D, Gao S, Christensen JG, Weiser MR. MET activation mediates resistance to lapatinib inhibition of HER2-amplified gastric cancer cells. Mol Cancer Ther. 11:660–9. doi: 10.1158/1535-7163.MCT-11-0754. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Liska D, Chen CT, Bachleitner-Hofmann T, Christensen JG, Weiser MR. HGF rescues colorectal cancer cells from EGFR inhibition via MET activation. Clin Cancer Res. 17:472–82. doi: 10.1158/1078-0432.CCR-10-0568. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Benedettini E, Sholl LM, Peyton M, et al. Met activation in non-small cell lung cancer is associated with de novo resistance to EGFR inhibitors and the development of brain metastasis. Am J Pathol. 177:415–23. doi: 10.2353/ajpath.2010.090863. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Agarwal S, Zerillo C, Kolmakova J, et al. Association of constitutively activated hepatocyte growth factor receptor (Met) with resistance to a dual EGFR/Her2 inhibitor in non-small-cell lung cancer cells. Br J Cancer. 2009;100:941–9. doi: 10.1038/sj.bjc.6604937. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Engelman JA, Zejnullahu K, Mitsudomi T, et al. MET amplification leads to gefitinib resistance in lung cancer by activating ERBB3 signaling. Science. 2007;316:1039–43. doi: 10.1126/science.1141478. [DOI] [PubMed] [Google Scholar]
- 33.Hanken H, Gaudin R, Grobe A, et al. Her2 expression and gene amplification is rarely detectable in patients with oral squamous cell carcinomas. J Oral Pathol Med. 43:304–8. doi: 10.1111/jop.12173. [DOI] [PubMed] [Google Scholar]
- 34.Citri A, Yarden Y. EGF-ERBB signalling: towards the systems level. Nat Rev Mol Cell Biol. 2006;7:505–16. doi: 10.1038/nrm1962. [DOI] [PubMed] [Google Scholar]
- 35.Sharafinski ME, Ferris RL, Ferrone S, Grandis JR. Epidermal growth factor receptor targeted therapy of squamous cell carcinoma of the head and neck. Head Neck. 32:1412–21. doi: 10.1002/hed.21365. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.