A Novel HSP90 Inhibitor–Drug Conjugate to SN38 Is Highly Effective in Small Cell Lung Cancer

Of the one in four deaths occurring annually due to cancer in the United States, approximately 16% (160,000 patients) are due to lung cancer (1). Small cell lung cancer (SCLC) represents approximately 12% to 14% of the total lung cancer population, affecting more than 30,000 patients annually, and has a particularly poor prognosis (2). Postdiagnosis median survival is 15 to 20 months for patients diagnosed with limited stage disease and 9 to 12 months for patients with extensive stage (ES) disease. In contrast to other tumor types, such as non–small cell lung cancer (3), SCLC typically is not characterized by kinase-activating mutations and other changes in targetable signaling pathways; rather, characteristic features include inactivation of the tumor suppressors TP53 and RB1 and changes limiting NOTCH and altering TP73 function (4, 5). Compatible with this profile, targeted therapies have not proven effective to date in SCLC. Standard front-line therapies include a combination of a platinum agent and etoposide or irinotecan/topotecan (6), and tumors typically recur rapidly and are resistant to second-line therapies. One critical barrier to effective treatment of SCLC tumors with cytotoxic agents has been the inability to concentrate these drugs in the tumor at sufficient levels to achieve therapeutic benefit without simultaneously inducing untenable degrees of toxicity in normal tissues. Although a great deal of effort has been devoted to developing strategies to concentrate cytotoxic agents within tumors (7), so far, these approaches have met with equivocal success.

STA-8666 (8) is a recently described tripartite molecule designed to address the issue of targeted delivery of cytotoxic agents to the tumor by using HSP90 as a tumor-targeting agent. As a result of stresses existing within tumors and the tumor microenvironment, HSP90 is both highly overexpressed and in an activated configuration in tumors relative to normal tissue (9, 10). STA-8666 is comprised of an HSP90-targeting moiety fused via a cleavable carbamate linker to the cytotoxic compound SN-38 [the active metabolite of irinotecan (11)], which inhibits topoisomerase I to induce double- and single-strand DNA breaks. As with irinotecan, carboxylesterase (CES) activity in normal tissues (primarily liver) and tumors cleaves the linker region, releasing SN-38 over an extended period (8). Hence, STA-8666 has the potential for greater antitumor activity than other SN-38 delivery vehicles because of its ability to be concentrated in tumor tissue, where intratumoral cleavage provides high, selective SN-38 exposure. Such an approach has theoretical advantages versus other targeting strategies, such as antibody–drug conjugates, in that low molecular weight (880 Da) STA-8666 does not require a cell-surface antigen for binding and active endocytosis, while the abundance of HSP90 in tumor cells obviates the requirement for a cytotoxic component active at extremely low concentrations.

The first study of STA-8666 reported exceptional biological activity of this compound against xenograft models of pancreatic and breast cancer and demonstrated activity was dependent on the ability of this compound to interact with HSP90 (8). In this study, we therefore evaluated STA-8666 in SCLC xenografts, benchmarking against standard first- and second-line therapies, determining whether tumors resistant to such therapies were responsive to STA-8666, and performing parallel mechanistic analysis. The extraordinary potency of STA-8666 in SCLC patient-derived xenograft (PDX) and cell line models, coupled with the identification of response biomarkers, provides significant preclinical support for the development of this agent toward clinical trials for SCLC.

The NCI-H69 human SCLC cell line (12), obtained from the ATCC, and the NCI-H157 (a gift from J. Kurie, Department of Thoracic Head and Neck Medical Oncology, University of Texas, MD Anderson Cancer Center, Houston, TX, USA, 77030) have been described. Cell lines were sent for authentication to IDEXX Bioresearch. Short tandem repeat analysis using the Promega CELL ID System (8 STR markers + amelogenin) was performed and verified that the genetic profile of the sample matches the known profile of the cell line. The samples were confirmed to be of human origin, and no mammalian interspecies contamination was detected.

The LX-36 (also known as CTG-0199 and BML-4) PDX model was derived at Fox Chase Cancer Center (Philadelphia, PA; see Supplementary Methods) and was used at passage 4. The genotype of mutations relevant to common mutations in SCLC was determined by exome sequencing performed by Ambry Genetics.

For xenograft analysis of the NCI-H69 and SCLC1 cell lines, 1 × 10⁷ cells were introduced into the flank of 6- to 8-week-old female C-B17.SCID mice. For the LX-36 model, tumors were minced on ice to 1- to 2-mm fragments, mixed 1:1 in RPMI/Matrigel Basement Membrane Matrix (#354234; BD Biosciences), and implanted subcutaneously in both flanks of 25 C-B17.SCID mice. Mice were monitored twice a week. Tumor volumes (V) were calculated by caliper measurement of the width (W), length (L), and thickness (T) of tumors using the following formula: V = 0.5236 × (L × W × T). Mice with tumor size in the range of 90 to 250 mm³ (∼day 24, and 60% of total mice; average size 150 mm³) were selected and randomized into treatment groups for treatment with vehicle, STA-8666, irinotecan, carboplatin, etoposide, topotecan, or ganetespib (details in Supplementary Methods). Mice with tumors >1,500 mm³ or experiencing 10% weight loss or with appearance of distress were switched from original treatment to 150 mg/kg STA-8666, as detailed in Results. All mice were euthanized by CO₂ inhalation at 120 days after commencement of dosing, or in cases of distress, or >10% weight loss. Tumors were excised and divided for formalin fixation and paraffin embedding (FFPE) or flash frozen and stored at −80°C.

The NCI-H69 cell line was used for assessment of pharmacokinetics and intratumoral response following transient dosing. Xenografts were established as described above, except tumors were allowed to grow to 1,000 mm³ before commencing dosing. For signaling, when tumors reached 600 mm³, mice were injected with vehicle, STA-8666, or irinotecan, then euthanized at times indicated after dosing as noted in Results, with tumor tissue collected, divided, and processed by FFPE or flash freezing. For kinome analysis, tumor tissues were weighed and homogenized in a 3× volume of PBS containing 40 mg/mL NaF, extracted by protein precipitation and analyzed using a LC/MS-MS with a Waters Symmetry Shield RP18 column (5 μm, 2.1 × 100 mm).

Tumor xenografts and plasma samples were collected and snap frozen 72 hours after a single injection of 150 mg/kg STA-8666 or 60 mg/kg irinotecan and sent for further analysis to Synta Pharmaceuticals. Plasma and tumor samples were received frozen in dry ice and stored at −80°C. Prior to analysis, tumor samples were weighed and homogenized in 3× volumes of PBS containing 40 mg/mL sodium fluoride, an esterase inhibitor. Homogenized tissue and plasma samples were extracted by protein precipitation with 0.5/95.5 (v/v) formic acid/acetonitrile containing stable isotope–labeled internal standards for STA-8666, STA-12-8663, and SN-38. An Agilent 1100 high pressure liquid chromatography pump was coupled to a Waters Symmetry Shield RP18 column (100 × 2.1 mm, 5 μm). The analytes were eluted at a flow rate of 0.5 mL/minute using a gradient mobile phase composed of 0.1% formic acid in water (solvent A) and 0.1% formic acid in acetonitrile (solvent B). The eluent was introduced via electrospray ionization directly into an AB Sciex API 4000 triple quadrupole mass spectrometer operated in the positive ion mode. Total run time was 5 minutes per sample.

Pathology specimens were stained with hematoxylin and eosin (H&E; Sigma-Aldrich). Tumor sections were immunostained with antibodies to cleaved caspase-3 (Cell Signaling Technology), γH2AX (EMD Millipore), HSP70 (Novus Biologicals), HSP90 (Cell Signaling Technology), and Ki67 (Dako). All analyses were performed by a board-certified pathologist (K.Q. Cai) blind to sample identity. Quantification of signal in immunostained slides was performed using an Aperio ScanScope CS scanner (Aperio) and Vectra Automated Quantitative Pathology Imaging System (PerkinElmer; see Supplementary Methods).

To analyze the expression levels of individual proteins, tumor tissue and cells were lysed in T-PER lysis buffer (Thermo Scientific) and CelLytic MT Cell Lysis Reagent (Sigma-Aldrich). Protein concentrations of the resulting lysates were measured using the Pierce BCA Protein Assay Kit (Thermo Scientific). Western blotting was performed using standard procedures, and blots developed by chemiluminescence using Luminata Western HRP substrates (Classico, Crescendo and Forte, EMD Millipore) and Immun-Star AP Substrate (Bio-Rad Laboratories).

Primary antibodies were used in 1:1,000 dilution (if not indicated differently) and included anti-PARP (#9542), anti-phospho CHK1 Ser345 (#2348), anti-CHK1 (#2360), anti-phospho cyclin B1 Ser133 (#4133), anti-cyclin B1 (#4135), anti-phospho-CDK1 Tyr15 (#9111; all from Cell Signaling Technology); anti-HSP70 (monoclonal, mouse, #ADI-SPA-820, Enzo) or anti-HSP90α (#ADI-SPS-771, Enzo and #4877, Cell Signaling Technology); anti-CDK1 (#sc-54, Santa Cruz Biotechnology); anti-KAP1 Ser824 (ab70369), anti-KAP1 (ab56587), anti-GAPDH (ab9482 and ab8245, 1:10,000; from Abcam); and anti-vinculin (mouse, monoclonal hVIN-1, #V9131, Sigma-Aldrich). Secondary anti-mouse and anti-rabbit HRP-conjugated antibodies (GE Healthcare) were used at a dilution of 1:10,000, and secondary anti-mouse and anti-rabbit AP-conjugated antibodies (Jackson ImmunoResearch Labs) were used at a dilution of 1:5,000. Quantification of signals on Western blots was done using the NIH ImageJ Imaging and Processing Analysis Software (13), with signaling intensity normalized to loading control. IRDye800CW goat anti-mouse and IRDye 680 RD goat anti-rabbit secondary antibodies were used at 1:20,000 dilution (LI-COR). Membranes were imaged and quantified in separate channels (700 and 800 nm) in the Odyssey Infrared Imaging System using Odyssey V3.0 software (LI-COR).

Flash-frozen tumors collected from mice treated for 48 hours with either vehicle control, 150 mg/kg and 50 mg/kg HDC, or 60 mg/kg irinotecan (n = 3/group) were randomized into three experimental replicate groups consisting of a single tumor from each of the three treatment conditions for processing and subsequent quantitative LC/MS analysis. Dr. Gary Johnson of the University of North Carolina at Chapel Hill (Chapel Hill, NC) kindly provided the inhibitor-conjugated beads VI16832 and PP58. Multiplexed inhibitor beads (MIB) preparation and chromatography was performed as described in ref. 14 and detailed in Supplementary Methods.

H69 and H157 cells treated for 24, 48, and 72 hours with 100 nmol/L of STA-8666 or irinotecan were collected, pelleted by centrifugation at 1,500 rpm for 4 minutes, and suspended in 0.5 mL PBS. Cells were subsequently fixed and stained for phosphorylated (γ) H2AX as described in ref. 15. Cells were incubated overnight at 4°C in anti-γH2AX primary antibodies (1:200; mouse monoclonal, Millipore Upstate) and 1 hour at room temperature with FITC-tagged secondary antibodies [1:500; goat anti-mouse IgG (H+L) secondary antibody, Alexa Fluor 488 conjugate, Thermo Scientific]. Cells were pelleted and resuspended in 300 μL of propidium iodide solution (PI/RNase Staining Buffer, BD Pharmingen). Following incubation at room temperature for 30 minutes, cell-cycle distribution and γH2AX expression were analyzed using FACS (Becton Dickinson).

For Figs. 1G and 2B, we used Fisher exact tests for comparison of complete response rates between groups. For the other figures, we generally used Wilcoxon rank-sum tests for pairwise comparisons, unless otherwise noted. Analyses were performed by using STATA versions 12 and 13 and Prism 6 (GraphPad Software).

We first examined the effectiveness of weekly STA-8666 in comparison with vehicle or irinotecan in controlling xenograft growth for the NCI-H69 (12) SCLC cell line in SCID mice (Fig. 1 and Supplementary Fig. S1). For this, we explored a dose range below the previously identified MTD of 200 mg/kg (16) to establish efficacy in the context of a clear therapeutic window. For mice dosed with STA-8666 at 150 mg/kg, tumors regressed below detectability after 3 initial doses (Fig. 1A). After the third dose, observations continued without treatment. No recurrence was observed for 3 of 6 mice in STA 150 mg/kg group over 3 months; recurrent tumors in 3 of 6 mice remained sensitive to redosing and were eliminated with 3 further weekly doses (Fig. 1A). Importantly, no weight loss or sign of distress was seen in mice receiving 150 mg/kg STA-8666 or irinotecan at 60 mg/kg (Supplementary Fig. S1A).

In contrast, tumors grew very rapidly in mice dosed with vehicle, approaching volumes that required humane euthanasia within 3 weeks (Fig. 1B). However, crossing these animals onto STA-8666 treatment led to rapid, complete responses (Fig. 1C). By comparison, continuous weekly treatment of animals with the SN-38 prodrug irinotecan at 60 mg/kg, representing a concentration equimolar for the SN-38 moiety of STA-8666 at 100 mg/kg, also resulted in stable disease (Fig. 1D and Supplementary Fig. S1A). Dosing with irinotecan at 100 mg/kg, a level equimolar for SN-38 from STA-8666 at 150 mg/kg, exceeded the MTD for this compound, based on initial safety studies.

Dosing with STA-8666 at lower levels (100, comparable with irinotecan 60 mg/kg; and 50 mg/kg) was less effective than dosing STA-8666 at 150 mg/kg, with 100 mg/kg causing transient complete regression of tumors and 50 mg/kg controlling tumor growth for up to 3 months (Fig. 1E and Supplementary Fig. S1B and S1C). However, these dose levels required continuous weekly administration of STA-8666 to maintain stable disease. In additional benchmarking experiments to compare compounds currently used for SCLC, topotecan (12.5 mg/kg weekly) and etoposide (8 mg/kg, 3 days/week) used at concentrations at MTD induced limited tumor control in the first 2 weeks of dosing, but each caused significant side effects, including rapid weight loss, forcing discontinuation of treatment (Fig. 1E and Supplementary Fig. S1D and S1E). Similar to a recent report (17), inhibition of HSP90 with ganetespib only partially inhibited the growth of SCLC xenografts, and mice treated with this agent gradually lost weight as tumors progressed (Fig. 1E and Supplementary Fig. S1F). Carboplatin (30 mg/kg) caused significant weight loss and only modestly slowed tumor growth even with continuous dosing, with 4 of 5 mice showing early resistance (Fig. 1E and Supplementary Fig. S1G). However, combination of 30 mg/kg carboplatin with STA-8666 at 50 mg/kg was highly effective in controlling tumor growth, as after three doses, 11 of 11 tumors shrunk below detection limits, and the combination was well tolerated (Fig. 1E and Supplementary Fig. S1H). This response was durable for at least 120 days in 6 of 11 mice.

Mice previously treated with vehicle, 50 mg/kg STA-8666, carboplatin, topotecan, etoposide, or 30 mg/kg carboplatin + 50 mg/kg STA-8666 that had progressed, had stable disease for 6 to 8 weeks, or were unable to tolerate treatment, were then administered 3 doses of STA-8666 at 150 mg/kg (Fig. 1F and Supplementary Fig. S1). Importantly, in all cases except prior dosing with STA-8666 at 50 mg/kg, this resulted in rapid and sustained elimination of tumors, cessation of negative side effects, and regaining of lost weight (Supplementary Fig. S1). In summary, statistical analysis of results comparing efficacy of all treatments by 3 weeks of dosing showed much higher efficacy of STA-8666 at 150 and 100 mg/kg, and STA-8666 at 50 mg/kg combined with carboplatin at 30 mg/kg, versus all other treatments (Fig. 1G).

As an independent model of SCLC, we used the PDX LX-36 (Supplementary Table S1) and compared STA-8666 (150 and 75 mg/kg) to vehicle, irinotecan (60 mg/kg), and the combination of etoposide (4 mg/kg, days 1, 2, 3) plus carboplatin (30 mg/kg; Fig. 2 and Supplementary Fig. S2). As with the NCI-H69 model, three doses of 150 mg/kg STA-8666 eliminated tumors in 5 of 5 animals, and these responses remained durable until at least 80 days from the start of treatment in 4 of 5 animals. In the single animal in which a tumor recurred, it was eliminated with another 3-week cycle of treatment (Fig. 2A), resulting in a statistically significant outcome from vehicle-treated animals (Fig. 2B). Vehicle-treated animals progressed rapidly (Fig. 2C), but substitution of treatment with STA-8666 2 to 3 days prior to ethically required euthanasia durably eliminated large tumors, as in NCI-H69 xenografts (Fig. 2D). Over a month of continuous weekly dosing, 75 mg/kg STA-8666 caused tumor regression or stable disease; subsequent substitution of 150 mg/kg caused transient regression, but 2 of 3 tumors ultimately grew uncontrollably (Fig. 2E). Neither irinotecan at its MTD nor etoposide/carboplatin effectively controlled tumor growth in this model; however, administration of STA-8666 at 150 mg/kg was again well tolerated and effective in reducing or eliminating detectable tumors in mice unsuccessfully treated with the other compounds (Fig. 2F and G).

The high efficacy and low toxicity of STA-8666 in tumor xenografts is compatible with higher retention of the SN-38 moiety in the tumor. We analyzed retention of compounds in plasma versus tumors of mice with established H69 xenografts treated 72 hours previously with either 150 mg/kg STA-8666 or 60 mg/kg irinotecan. The prodrug irinotecan and its metabolite, SN-38, were below quantifiable limits in irinotecan-treated animals, suggesting rapid clearance. STA-8666 was concentrated 46-fold in tumor (7.5 nmol/g tissue) versus plasma (163 nmol/L), whereas its cleaved SN-38 metabolite was concentrated 10-fold (0.56 nmol/g tumor vs. 0.05 nmol/L in plasma; Fig. 3A). The high tumor concentration of STA-8666 in mice treated with STA-8666 suggested that at this time point, much of the compound was uncleaved, leaving a residual pool for continual SN-38 release. Notably, the levels of SN-38 released from STA-8666 were approximately 4-fold lower than intact STA-8666, suggesting that SN-38 is actively being cleared from tumor cells, as inferred for irinotecan.

We performed quantitative histopathologic analysis of tissues harvested from mice 72 hours after a single dose of 150 mg/kg STA-8666, 60 mg/kg irinotecan, or vehicle (equivalent to the time point for pharmacokinetic analysis) or 48 hours after a second weekly dose of compound, a time at which STA-8666–treated tumors began to significantly shrink (Fig. 3B–E and Supplementary Fig. S3). At both time points, the necrotic area of STA-8666–treated tumors exceeded that of irinotecan- and vehicle-treated tumors (59% vs. 41% and 27% at the first time point; and 44% vs. 29% or 39% at the second). By the second time point, STA-8666 had significantly reduced total tumor area versus vehicle or irinotecan [36 vs. 64 mm² for vehicle (P = 0.007) or 55 mm² for irinotecan (P = 0.00013; Fig. 3B–D and Supplementary Fig. S3)]. In addition, tumor size reduction was more homogeneous with STA-8666 than irinotecan (P = 0.003 at the first time point and P = 0.054 at the later time point). By the early time points, both irinotecan and STA-8666 modestly elevated expression of the DNA damage marker phosphorylated (p) H2AX and the apoptotic marker, cleaved caspase-3. By the later time point, both pH2AX and cleaved caspase-3 were strongly elevated by STA-8666 relative to vehicle (5-fold and 13-fold) or irinotecan (2.3-fold and 10.5-fold) in nonnecrotic areas of the tumor (Fig. 3E and Supplementary Fig. S3B).

Levels of HSP90 were generally consistent between all treatment groups at both time points, with a slight reduction in expression associated with high treatment levels of STA-8666 at the early time point. No statistically significant change in HSP70 expression was observed, reflecting the biologically insignificant HSP90-inhibitory activity of STA8666 (8). Interestingly, analysis of the small number of tumors treated with lower doses of STA-8666 that developed resistance to this compound (data in Supplementary Fig. S1B and S1C) showed a similar profile of HSP70 and HSP90 expression to transiently treated tumors, except that 2 of 3 resistant tumors had barely detectable levels of pH2AX (Supplementary Fig. S4), potentially suggesting greater capacity to efflux the compound. Together, these results indicate that the primary biological activity of STA-8666 is through prolonged SN-38 exposure provided by the selective retention of STA-8666 in the tumor and superior to that of irinotecan.

Cellular exposure to SN-38 is characteristically associated with DNA damage and G₂–M checkpoints and previously reported as maximal at 24 to 48 hours following cell exposure to SN-38 (Fig. 4A; refs. 18, 19). We hypothesized the targeting activity of STA-8666 and associated greater retention of SN-38 in tumors would result in a stronger and sustained imposition of DNA damage, and stimulus to arrest, than would irinotecan. Supporting this idea, tumors harvested from mice 24, 48, or 72 hours after STA-8666 treatment showed a more rapid, robust, and sustained induction of phS⁸²⁴-KAP1 (Fig. 4B), a product of damage-associated ATM phosphorylation, compared with irinotecan (20). Similar results were observed with the checkpoint-associated phS³⁴⁵-CHK1 protein (Fig. 4C). STA-8666 caused marked elevation of phY¹⁵-CDK1 (Fig. 4D) and downregulation of cyclin B1 (CCNB1) and PLK1-phosphorylated phS¹³³-CCNB1 (Fig. 4E) (21), compatible with imposition of a strong G₂–M checkpoint arrest by STA-8666, with this peaking at 24 to 48 hours, as described previously (18, 19), and more marked for 150 mg/kg STA-8666 than observed with irinotecan. Cleavage of PARP1 was strongly increased in all drug-treated tumors, with this indicator of apoptosis peaking at 72 hours and much higher in tumors treated with STA-8666 than with irinotecan (21-fold for 150 mg/kg STA-8666; Fig. 4F). Aside from a transient minor elevation of HSP70, expression of HSP90 and HSP70 was not affected by either drug treatment (Fig. 4G).

To further support this analysis, we used a recently described MIB (14, 22) technique coupled with mass spectrometry to profile the intratumoral activity of 148 kinases 48 hours after dosing 3 mice bearing NCI-H69 tumors with irinotecan (60 mg/kg), STA-8666 (50 mg/kg), or vehicle (Fig. 5A). STA-8666 durably induced a number of kinases associated with onset of G₂–M arrest and DNA-damage checkpoints, whereas irinotecan treatment did not; rather, irinotecan-treated tumors showed modest depression of many kinases relative to vehicle treatment, potentially due to the weaker, more transient levels of drug in the tumor. In summary, these data suggested a more robust activity of STA-8666 in inducing G₂–M arrest, potentially because of the more durable concentration and retention of the compound. To further test this idea, we compared the ability of STA-8666 and irinotecan treatment to induce cell-cycle arrest in vitro (Fig. 5B). At isomolar concentrations of 100 nmol/L, STA-8666 induced a more significant accumulation of cells in G₂–M than did irinotecan in both NCI-H69 and NCI-H157 lung cancer cells, with differences appearing 24 to 48 hours after addition of drug. Paralleling these results, flow cytometry analysis of treated cells indicated that STA-8666 also more effectively induced γH2AX at similar time points (Fig. 5C). On the basis of Western blot analysis, STA-8666 also induced phS⁸²⁴-KAP1 and phY¹⁵-CDK1, at earlier time points and to a greater degree than irinotecan, in a dose-responsive manner (Fig. 5D and E and Supplementary Fig. S5A and S5B). Subsequently, PARP cleavage was elevated, with the greatest effect seen with 100 nmol/L STA-8666 (Fig. 5F and Supplementary Fig. S5). HSP90 and HSP70 levels were unaffected (Fig. 5G and H and Supplementary Fig. S5).

Our data for the first time demonstrate potential clinical value for application of STA-8666 in the setting of SCLC. Our results differentiate STA-8666 from irinotecan in both the intensity of the DNA damage and cell-cycle checkpoint response, as well as the induction of apoptosis and necrosis in human tumors. These results are compatible with pharmacokinetic results indicating delivery of high levels of SN-38 directly to target tumor tissue and superior retention over time, which is important in tumors with intrinsic resistance to therapeutics, such as SCLC (23). Systemic treatment for SCLC has not significantly improved for decades (2), while 5-year survival remains at <7% overall; for the 60% to 70% of patients diagnosed with ES SCLC, 2-year survival is <5% (24). Factors contributing to these abysmal figures include the failure of early detection approaches to effectively identify rapidly dividing SCLC tumors before progression to the ES stage, and the characteristic mutational profile of SCLC, which has differed from many other tumor types in failing to suggest strategies for application of targeted therapeutic agents (2).

In this context, an HSP90-targeted strategy is potentially particularly valuable, as the rapid proliferation rate of the tumors enforces dependence on HSP90 expression. Indeed, the prior treatment of SCLC with other cytotoxic agents or radiation may contribute to the vulnerability to STA-8666, as such treatments would increase cellular dependence on heat shock and other stress response systems (25). The initial report of STA-8666 demonstrated that although an HSP90-binding capacity was important for functional activity, HSP90 inhibition per se was insufficient to limit tumor growth (8); although we cannot rule out a minor contribution of HSP90-inhibitory activity in the observed tumor control, the data in this study confirm that the HSP90 inhibitor ganetespib was ineffective in controlling SCLC xenograft growth, and we did not observe elevation of the heat shock response.

In this study, we also found that prior treatment with a number of agents currently used for front-line therapy for SCLC did not result in resistance to STA-8666; rather, rapid tumor shrinkage was observed after substitution of treatment with 150 mg/kg STA-8666, and responses were typically durable. In addition, the combination of low-dose (50 mg/kg) STA-8666 with carboplatin was extremely effective. In this work, the only prior treatment that led to resistance to 150 mg/kg of STA-8666 was the same compound used at lower doses. This is compatible with the interpretation that these lower doses are not sufficient for complete eradication of tumors in the in vivo microenvironment, in spite of the large concentration of STA-8666 and SN-38 relative to plasma. Given the dose-limiting toxicity of untargeted SN-38, it has previously been impossible to perform analysis of curative doses for SCLC, which our studies suggest must likely fall in the range of 0.6 nmol/L/g tumor tissue. For mice previously treated with high doses of STA-8666, responses were either cure or extending through the 4 months of observation, or tumor recurrence, followed by subsequent remission. Subsequent experiments prior to clinical development of STA-8666 should address long-term consequences of relapse, whether tumors remain susceptible to cycles of redosing, and what mechanisms contribute to resistance to this compound.

Over the past several decades, advances in cancer treatment have divided tumors into classes for which very significant gains have been made, yielding regimens that are curative or suitable for chronic maintenance, versus other classes for which treatments remain largely ineffective. Taken in sum, these results strongly support the further development of STA-8666 for clinical evaluation in limited SCLC or ES SCLC in either the first- or second-line setting.

Y. Boumber reports receiving a commercial research grant from Synta Pharmaceuticals and is a consultant/advisory board member for Bristol-Myers Squibb, Clovis, and Novartis. No potential conflicts of interest were disclosed by the other authors.

Conception and design: A.V. Gaponova, A.S. Nikonova, Y. Boumber, E.A. Golemis

Development of methodology: A.V. Gaponova, K.Q. Cai

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A.V. Gaponova, A.S. Nikonova, A.Y. Deneka, M.C. Kopp, A.E. Kudinov, N. Skobeleva, V. Khazak, K.Q. Cai, K.E. Duncan, J.S. Duncan, Y. Boumber

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A.V. Gaponova, A.S. Nikonova, A.Y. Deneka, A.E. Kudinov, V. Khazak, L.S. Ogawa, K.Q. Cai, K.E. Duncan, J.S. Duncan, B.L. Egleston

Writing, review, and/or revision of the manuscript: A.V. Gaponova, A.S. Nikonova, A.Y. Deneka, B.L. Egleston, D.A. Proia, Y. Boumber, E.A. Golemis

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A.V. Gaponova, M.C. Kopp, V. Khazak, D.A. Proia

Study supervision: A.V. Gaponova, D.A. Proia, Y. Boumber, E.A. Golemis

We thank Vladimir Khazak and Champions Oncology for providing the CTG0199 (LX-36) PDX model and Synta Pharmaceuticals for providing ganetespib and STA-8666 for these studies.

The authors were supported by R21 CA181287 and R01 CA063366 (to E.A. Golemis); the Lung Cancer Research Foundation, the American Hellenic Educational Progressive Association (AHEPA), and the American Cancer Society IRG program (to Y. Boumer); by Russian Science Foundation project 15-15-20032 (to A.Y. Deneka); and the NIH core grant CA006927 (to Fox Chase Cancer Center).

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.