3. Discussion
This study demonstrated that patients in randomized control trials receiving Hsp90 inhibitors had a statistically significant increase in reported abdominal pain versus control. Back pain and headache showed a trend towards increased incidence. Peripheral neuropathy, myalgia, pain in the extremities, and arthralgia were also commonly reported in the patients treated with Hsp90 inhibitors. This study is the first to directly compare the effect of treatment with Hsp90 inhibitors versus controls on types of pain reported as adverse events.
Several Hsp90 inhibitors have been evaluated for their impact on tumor progression. This first generation of Hsp90 inhibitors includes 17-AAG, 17-DMAG, and IPI-504 [
26]. They bind to the ATP-binding site at the N-terminal domain of Hsp90, resulting in the degradation of most client proteins [
3]. Those Hsp90 inhibitors showed activities to reduce the tumor growth in the clinical trials. Meanwhile, patients treated with first generation Hsp90 inhibitors were subject to severe hepatic and ocular toxicity [
27]. In an effort to diminish these adverse effects, the second generation of Hsp90 inhibitors has been developed. This new generation targets the C-terminal domain of Hsp90, such as onalespib (AT13387), ganetespib (STA-9090), luminespib (AUY922), SNX-5422, BIIB-021, PU-H71, XL-888, HSP990, KW-2478, Debio-0932, KU-32, TAS-116, and dacinostat (LAQ-824). These drugs have successfully decreased the incidence of hepatic and ocular toxicity [
26]. However, some severe pain-related adverse effects have been reported in trials testing either the first or second generations of Hsp90 inhibitors. The current process of discovering Hsp90 inhibitors is targeting specific isoforms or co-chaperones of Hsp90, which might be helpful for eliminating such side effects shown in the patients.
Hsp90 has been demonstrated to be involved in the sensitization of pain. Levels of Hsp90 messenger RNA were increased in dorsal root ganglia in an arthritis rat model, and administration of 17-DMAG reduced monoarthritis-induced mechanical allodynia [
18]. Similarly, chronic constriction injury-induced allodynia was suppressed with systemic administration of 17-AAG and intrathecal administration of 17-DMAG in mice [
19]. KU-32 is a novobiocin analogue Hsp90 inhibitor being investigated for the treatment of diabetic neuropathy. In preclinical mouse studies, treatment with this Hsp90 inhibitor demonstrated slowed disease progression through the motor and sensory nerve conduction velocity [
20]. Additionally, the administration of KU-32 reversed disease progression markers, including motor and sensory nerve conduction velocity, mechanical and thermal sensitivity, and intra-epidermal nerve fiber density [
20,
21]. These preclinical findings suggest that Hsp90 inhibitors reduce the sensitization of different types of pain, such as neuropathic pain. In this study, the meta-analysis found that treatment with Hsp90 inhibitors increased abdominal pain and back pain (in trend), but showed no significant association between treatment with Hsp90 inhibitors and other types of pain in clinical trials. These differences may be due to the pain investigated in the preclinical models that were not cancer related. Another reason may be related to the activity of Hsp90 inhibitors on pain drugs, such as opioids. Previous studies showed that inhibition of Hsp90 in the mouse brain could diminish the opioid-induced signaling and behavior [
28,
29]. The increased risk of pain by Hsp90 inhibitors could relate to the attenuation of pain management in the patients. Unfortunately, none of the randomized controlled trials reported how the pain was managed in those patients.
Several limitations should be considered when the meta-analysis results are interpreted. First, the number of published studies that were eligible for analysis was relatively small. The majority of the clinical trials were phase I or II without control groups. However, heterogeneity was tested, and variations were controlled when the studies were combined for analysis. Second, indications and control groups in each study varied, which increased the level of heterogeneity of the publications. The random-effect model was applied to address the heterogeneity in the analysis. Due to the low incidence of arthralgia and myalgia reported in some studies, the wide 95% CIs could bias the results. Third, based on available information reported in clinical trials, the pain events were classified by anatomical area. Since the different types of underlying malignancy could affect the type of pain, assessing the pain associated with the Hsp90 inhibitor by underlying malignancy would enhance the knowledge of pain-related side effects in Hsp90 inhibitor development. In addition, pain outcomes were reported by patients in the trials. Whether patients were informed of pain as a potential side effect in the trial could influence patients’ perceptions and confound pain outcome measures. However, such information was not available in the publications. Finally, some trials were terminated without results. Thus, publication bias could influence the conclusion.