HSP90 Inhibitors and GIST

Sebastian Bauer, MD and Jonathan Fletcher, MD

Dr.  Bauer and Dr. Fletcher kindly shared with several GIST groups this summary about Heat Shock Protein 90 as a target in treating GIST, and the drugs under development to inhibit HSP90.

Dr. Bauer is a medical oncologist at the West German Cancer Center, in Essen, Germany.  He previously completed a post-doctoral fellowship in the lab of Dr. Fletcher at Brigham and Women’s Hospital.  In 2005 Dr. Bauer won the Young Investigator award at the Connective Tissue Oncology Society meeting for his work on heat shock protein 90 and its role in resistance of GIST to imatinib.  In 2006 he received a research award from the Liddy Shriver Sarcoma Initiative, with support from Team GIST (link to story).  He currently runs the soft tissue and bone sarcoma program at the West German Cancer center and works as a translational scientist in sarcoma research. Dr. Bauer regularly organizes a yearly sarcoma ride in Essen to support sarcoma research (see separate story).



Dr. Fletcher is an associate professor in the departments of Pathology and Pediatrics at Harvard Medical School and is an Associate Pathologist at Massachusetts General Hospital and at Brighamand Women’s Hospital.  He is also an oncologist at The Children’s Hospital in Boston.  Dr. Fletcher has been pursuing GIST research for many years and has published widely.  He has helped to train numerous GIST researchers as they have completed postdoctoral study in his lab, including Brian Rubin, MD, Anette Duensing, MD, and Sebastian Bauer, MD.  Dr. Fletcher’s lab has developed many of the GIST cell lines being used in preclinical research today.




HSP90 Inhibitors and GIST


Imatinib has revolutionized the treatment of unresectable GIST, with remarkable remissions in most patients, even in the face of widely metastatic disease;  the recent worldwide approvals of this agent in the post-surgical adjuvant setting also has promise to improve the outcomes of patients further. This clinical advance is all the more remarkable in that metastatic GIST was previously regarded as among the most untreatable, incurable forms of sarcoma. Nonetheless, as we have learned in the past 8 years, most GIST patients eventually develop resistance towards imatinib and that further improvements in our therapeutic approach are needed to achieve the ultimate goal of a cure for all.

One of the major mechanisms for imatinib resistance is the emergence of GIST cells that harbor secondary mutations of the KIT kinase domain (1). While primary mutations are usually found next to the cell membrane, kinase domain mutations are located in the section of KIT that transfers energy (in the form of phosphorylation) to its downstream signaling intermediates.  These secondary mutations have a major impact on the effectiveness of direct KIT inhibitors, such as Imatinib, in binding and inhibiting KIT and PDGFRA oncoproteins in GIST cells. So far, more than 10 different secondary mutations have been found in patients with imatinib-resistant GIST. Imatinib perfectly fits into the ATP-binding pocket, the docking station for the ATP-“battery packs”, in mutant KIT with primary mutations only. Secondary mutations still allow ATP to provide energy for KIT to function but decrease to various degrees the capacity of imatinib to compete with ATP.

Using the large inhibitor libraries used in drug development, several inhibitors were identified that do a better job than imatinib in inhibiting mutant KIT containing secondary mutations. However, none of these drugs (e.g. sunitinib, nilotinib, dasatinib or sorafenib) inhibits all mutations and especially mutations in the kinase-loop part of the kinase domain seem to be specifically difficult to inhibit. In addition, the efficacy of these second generation KIT inhibitors is hampered by the fact that different secondary mutations can occur within the same patient (2). This explains why we often see “mixed responses” in patients treated for imatinib-resistant GIST. While some metastases shrink, presumably with favorable secondary mutations, others continue to grow, despite treatment. A solution to this problem of genomic heterogeneity might either be a combination of several KIT inhibitors or a single drug that inhibits KIT regardless of the type of secondary mutation.

Mast cell disease or mastocytosis is a chronic hematological disease that is characterized by the presence of too many mast cells. As in GIST, the growth of these cells is often driven by constitutively activated KIT, which often harbors mutations in the KIT kinase-loop.  In 2004 researchers from the NCI showed, that a drug called 17-AAG, or 17-Allylamino-17-demethoxygeldanamycin, effectively inhibited KIT at low doses even in mast cells that contained the kinase-loop mutation D816V. Notably, this particular KIT mutation causes resistance to imatinib, in a subset of GIST patients.  It has been estimated that inhibition of the D816V mutant KIT by imatinib would presumably require doses of more than 10 grams per day, which cannot be given to patients.

Inspired by this work, we investigated 17-AAG as a potential KIT-inhibitor in GIST cells and found – similar to the aberrant mast cells – GIST cells harboring mutant KIT were highly sensitive to 17-AAG (3).  At the same time, a new water-soluble derivative of 17-AAG called IPI-504 (now known as retaspimycin)(4;5) was about to enter clinical trials. This fortunate availability of a clinically useful compound allowed the treatment of a first GIST patient by our clinical colleague, Dr. Demetri, only a few months later. The initial results were promising, and in short order a worldwide multicenter phase III trial (called the RING trial, for “Retaspimycin In GIST”) was begun in an attempt to evaluate rapidly the effectiveness of this therapeutic approach in patients with imatinib- and sunitinib-refractory GIST.

What is 17-AAG and why is HSP90 inhibition a therapeutic opportunity in GIST?

In 1970, researchers searching for new antibiotics first described a new bacteria strain (Streptomyces hygroscopicus var. geldanus) that they found in the soil of Kalamazoo, Michigan. These bacteria  produced an antibiotic which they called Geldanamycin(6). 17-AAG is a chemically-modified version of Geldanamycin which has turned out to be more chemically stable and exhibit a better therapeutic window. Initially, Geldanamycins were thought to be kinase inhibitors, but it later turned out, that their kinase inhibition effects were a consequence of inhibiting  a protein called HSP90, Heat Shock Protein 90.

HSP90 is one of the most abundant proteins in human cells and belongs to a broad family of proteins, known as “chaperones”.  Chaperones are vital in maintaining the normal functions of cells, and are especially needed after environmental stresses that induce protein damage. Experiments in which cells are stressed by heat exposure have demonstrated strong induction of chaperone synthesis, and for this reason many chaperones are known as “heat shock proteins”. However, this name is misleading, as almost every abrupt change in the cellular environment induces chaperones, and their cellular functions are multifaceted.

HSP90 is responsible for controlling the conformation, stability, activation, intracellular disposition, and disposal (proteolytic turnover) of
many important proteins that are involved in cell growth, differentiation, and survival. Transcription and translation of genes results in a string of amino acids that, left on their own, assume a 3-dimensional configuration (like a self-inflatable life-raft…). However, those proteins don’t configure optimally on their own, and rather assemble into something like an unkempt ball of wool, which would be identified as useless and be disposed into the cells’ trash can, by the “proteasome.”  Proteins, such as mutant KIT, that rely on the help of the HSP90 chaperone machinery, are called HSP90 clients, and HSP90 ensures that these clients are properly folded and protected from premature destruction.

Cancer researchers have been intrigued by HSP90 inhibitors for many years, since the list of HSP90 clients contains the “celebrity A-list” of oncogenes, such as the EGF-Receptor, which is commonly deregulated in many types of epithelial cancer. When used in test tubes, HSP90 inhibitors, such as 17-AAG selectively kill cancer cells but not normal cells at low doses. This phenomenon has been explained by the fact that cancer cells are addicted to the oncogenic HSP90 clients and normal cells are not. 17-AAG shows a 100-fold higher affinity to HSP90 in tumor cells compared to the HSP90 protein in normal control cells. This can be explained by the fact that tumor cells are usually permanently stressed by their high demands for nutrients and oxygen. Therefore, oncogenic clients are in permanent need of the actively supporting, ATP-consuming HSP90 complex, the so-called super-chaperone complex (7), which can be inhibited by ATP-competitive drugs, such as 17-AAG.

In theory this sounds like a miracle drug. However, the first 12 trials of 17-AAG in cancer patients involving more than 300 patients showed not a single objective clinical remission. This seeming discordance between laboratory concept and clinical reality can be explained, in part, by the fact that cancer cells in lab test tubes are fully exposed to the drug being tested, whereas cancer cells in a human can be protected from the same drug by various physiological barriers.  In the case of 17-AAG or other geldanamycin based drugs (ansamycins), clinical applications have been constrained by the challenging pharmacology of the drugs.  They need to be administered intravenously and due to their low water solubility have needed to be dissolved in a solvent (until recently) which on its own had substantial side effects. In addition, most of the patients treated in these clinical trials had been randomly selected and had usually been heavily pretreated for epithelial cancers, which in part might explain the poor response rates.

Unlike many epithelial cancers, GISTs are genetically rather stable and mutant forms of KIT/PDGFRA are the dominant oncogenic drivers, which explains why imatinib or sunitinib work so well in GIST. We have shown that HSP90 inhibition results not only in the inhibition but also the proteasomal degradation  of oncogenic KIT irrespective of the imatinib-resistance mutations present (3).

In theory, GIST should therefore represent an ideal disease model for HSP90 inhibitor strategies. And, indeed, an initial clinical trial of the HSP90 inhibitor IPI-504 has provided the crucial proof-of-concept that HSP90 works as a master-regulator of KIT in GIST cells. In 66% of patients a reduction of glucose metabolism as measured by PET-scans was seen and one-third of patients experienced a prolonged disease stabilization (4).

IPI-504 is a chemically-modified form of 17-AAG which is highly soluble in water, thereby overcoming one of the major pharmacologic problems of 17-AAG. Notably, the phase I and II trials showed that IPI-504 was generally well-tolerated. In these early-phase trials, most of the observed IPI-504 toxicities were Grade 1 or 2, which means they were very mild and mostly required only outpatient management and no hospitalization. Side effects related to IPI-504 that occurred in more than 5 patients were asymptomatic slowing of heart rate, mild diarrhea, mild nausea (30%) and vomiting (14%), which was very responsive to standard anti-emetics. Other side effects were fatigue (50%), infusion site pain, asymptomatic increased liver enzymes, muscle and joint pain (35%) and mild headache (35%; (8)). Moderate or severe side effects (Grade 3 or 4) were observed in few patients in the initial review of the clinical data from this early phase trial, and the large scale international phase III trial (RING) was designed on this basis.  Sadly, the RING trial was recently closed early due to concerns that the frequency of patients who had severe side effects in the phase III trial was higher than predicted, perhaps because the patients had more advanced GIST and less normal functional reserve. Although it is too early to speculate on the exact mechanisms for these IPI-504 associated toxicities, it will be crucial to determine whether severe toxicities are inevitable with HSP90 inhibition in subgroups of GIST patients, particularly those who have already been extensively treated with other GIST therapies, and who may be in fragile medical condition. Alternately, it is possible that certain of the toxicities seen in the RING trial are not inevitable consequences of HSP90 inhibition per se, and might be circumvented by use of other classes of HSP90 inhibitor drugs.  The data that have been collected in this recently closed phase III trial will hopefully help us all to understand how HSP90 inhibition might be developed in the safest possible way as a therapeutic strategy for patients with GIST.

Beyond the potential for severe toxicities, it is also clear that substantial progress must be made in determining the optimal dose and schedule of HSP90 inhibitors, for treatment of GIST.  IPI-504 has been administered i.v. twice weekly for two weeks with one week off. Comparing PET scans during treatment a significant glucose reduction was seen at the end of the second week but after 1 week off drug administration in the first cycle of study, glucose metabolism in the GISTs had returned to near-baseline levels. While IPI-504 is the most advanced HSP90 inhibitor currently in clinical trials, there are limitations, such as the inconvenient schedule (frequent iv-treatment) and the fact, that HSP90 may not be continuously inhibited. However, IPI-504 might be available in an oral form in the near future and several other synthetic, orally available HSP90 inhibitors are being developed by other pharmaceutical companies (Table 1). Future studies are needed to identify the most potent, safe and clinically useful HSP90 inhibitors for GIST. An urgent priority will be to determine the extent to which HSP90 can be truly shut down by HSP90 inhibitor drugs, and to determine what clinical efficacy and what side effects result when that pharmacologic aim is achieved.  Other crucial laboratory studies include development of HSP90 resistance models and testing of strategies to prevent or overcome resistance to HSP90 inhibitors.


HSP90 inhibitors under development.

Compounds /
Drug Names
Route Develop-
Radicicol . . . . .
. Iv Phase II .
(cremaphor and
Iv Phase II Kosan
IPI-504 Retaspimycin Iv Phase III
IPI-493 . oral Phase I Infinity
17-DMAG KOS-1022
Iv and
(oil in water
. Iv Phase I/II Biogen
Macbecin . n.k. preclinical Biotica
Pyrazoles Resorcinol CCT018159 . . . .
. VER-49009
and others
. . . .
. BIIB021
(CNF 2024)
. oral Phase I/II
. AT-13387 . n.k. Phase I in
solid tumors
. PF-04928473 . oral Phase I in
solid tumors
. STA-9090 . IV Phase II in
. AUY922 . . Phase I in
solid tumors



Reference List

(1)  Heinrich MC, Corless CL, Blanke CD, Demetri GD, Joensuu H, Roberts PJ et al. Molecular correlates of imatinib resistance in gastrointestinal stromal tumors. J Clin Oncol 2006; 24(29):4764-4774.

(2)  Wardelmann E, Merkelbach-Bruse S, Pauls K, Thomas N, Schildhaus HU, Heinicke T et al. Polyclonal evolution of multiple secondary KIT mutations in gastrointestinal stromal tumors under treatment with imatinib mesylate. Clin Cancer Res 2006; 12(6):1743-1749.

(3)  Bauer S, Yu LK, Demetri GD, Fletcher JA. Heat shock protein 90 inhibition in imatinib-resistant gastrointestinal stromal tumor. Cancer Res 2006; 66(18):9153-9161.

(4)  Demetri GD, George S, Morgan JA, Wagner A, Quigley MT, Polson K et al. Inhibition of the Heat Shock Protein 90 (Hsp90) chaperone with the novel agent IPI-504 to overcome resistance to tyrosine kinase inhibitors (TKIs) in metastatic GIST: Updated results of a phase I trial. J Clin Oncol 2007; 25(18S (June 20 Supplement)):10024.

(5)  Bauer S, Yu L, Read M, Normant E, Demetri GD, Fletcher JA. IPI504, a novel HSP90 inhibitor, causes inhibition and degradation of KIT in imatinibresistant GIST: rationale for therapeutic targeting in GIST. AACR- NCI-EORTC Molecular targets and cancer therapeutics, Philadelphia 2005 2005;C49.

(6)  DeBoer C, Meulman PA, Wnuk RJ, Peterson DH. Geldanamycin, a new antibiotic. J Antibiot (Tokyo) 1970; 23(9):442-447.

(7)  Kamal A, Thao L, Sensintaffar J, Zhang L, Boehm MF, Fritz LC et al. A high-affinity conformation of Hsp90 confers tumour selectivity on Hsp90 inhibitors. Nature 2003; 425(6956):407-410.

(8)  Wagner A, Morgan JA, Rosen LS, George S, Gordeon MS, Devine CM et al. Inhibition of heat shock protein 90 (Hsp90) with the novel agent IPI-504 in metastatic GIST following failure of tyrosine kinase inhibitors (TKIs) or other sarcomas: Clinical results from phase I trial. J Clin Oncol 2008; 26(6):a10503.