Skip to main content
Premium Trial:

Request an Annual Quote

Race for Better Mutant PI3Kα Inhibitors Hinges on Disputed Science

Enhanced scanning electron micrograph of cancer cells

NEW YORK – Clinical trial readouts expected in 2024 could make or break mutant PI3Kα inhibitors in development by multiple drugmakers that hope to find the right recipe to challenge Novartis' Piqray (alpelisib), the use of which in breast cancer is hampered by a significant incidence of hyperglycemia.

While most of the players in the field, including Genentech, Relay Therapeutics, Loxo Oncology, and Scorpion Therapeutics, are pursuing strategies that target mutant PI3Kα and spare wild-type PI3Kα, newcomer Totus Medicines is focusing on sparing off-target inhibition of PI3Kβ, a controversial strategy in direct opposition to prevailing beliefs about the mechanisms that drive PI3Kα-mutant cancers and the hyperglycemic side effects associated with these targeted drugs. The companies each claim their approach is supported by evidence in the published literature and from preclinical studies, but the optimal strategy to minimize hyperglycemia and maximize anti-tumor efficacy remains to be confirmed by ongoing clinical trials.

The PI3 kinase pathway is one of the most frequently altered signaling networks in cancer. PIK3CA, the gene that encodes the P110α isoform of PI3 kinase, also known as PI3Kα, is mutated in about 13 percent of cancers, making it the second-most mutated gene behind TP53, which is altered in 35 percent of cancers. In comparison, mutations in KRAS and BRAF, targets that are the focus of some of the most competitive drug development markets, are found in just 11 percent and 8 percent of cancers, respectively.

When it comes to PIK3CA mutations, plenty of drugmakers are trying to develop better agents that hinder these mutations' ability to activate the PI3 kinase pathway that helps tumors grow and spread. The complexities of this pathway, however, have proven difficult for drugmakers to unravel when it comes to understanding why inhibition of the pathway causes hyperglycemia.

When insulin binds to insulin receptors on the cell surface, proteins inside the cell form a complex with PI3 kinase. That activates the insulin signaling pathway and leads to the translocation of glucose into the cell by the glucose transporter protein GLUT4. PI3Kα is believed to mediate insulin signaling within the cell, and one popular theory says the consequence of therapeutically inhibiting PI3Kα from activating the insulin signaling pathway and promoting cellular uptake of glucose is high glucose levels in the bloodstream.

The roles of the various PI3K isoforms including α, β, δ, and γ have not been fully characterized, however. A study published in Cell Metabolism in 2010 that explored the effects of liver-specific knockout of the p110α subunit of PI3 kinase in mice has been influential in forming the prevailing hypothesis that inhibition of wild-type PI3Kα by mutant PI3Kα inhibitors causes hyperglycemia in patients. In that study, deletion of p110α resulted in 15 percent to 20 percent increases in glucose levels in mice and impaired glucose tolerance.

Additional support for that theory comes from Novartis' Phase III SOLAR-1 clinical trial. In that trial, the company showed that patients with advanced hormone receptor (HR)-positive, HER2-negative breast cancer harboring at least one of 11 PIK3CA mutations who received Piqray with the estrogen receptor antagonist fulvestrant had a median progression-free survival of 11 months compared to 5.7 months for those that received fulvestrant alone. The overall response rate was 35.7 percent versus 16.2 percent, and median overall survival was 39.3 months versus 31.4 months, respectively.

But for patients with wild-type PIK3CA in this study, median progression-free survival was 7.4 months on Piqray-fulvestrant compared to 5.6 months on fulvestrant, a nonstatistically significant difference. The rate of hyperglycemia across the PIK3CA-mutated and -wild type cohorts was 64.8 percent in the Piqray-fulvestrant arm, compared to 9.4 percent in the placebo-fulvestrant group.

Novartis first reported SOLAR-1 results in 2018, and the US Food and Drug Administration subsequently approved Piqray in 2019 as a second-line therapy for HR-positive, HER2-negative metastatic breast cancer in combination with fulvestrant for patients with PIK3CA mutations alongside Qiagen's Therascreen PIK3CA PCR-based assay as a companion diagnostic for identifying patients with PIK3CA-mutated tumors eligible for treatment. PIK3CA mutations occur in 28 percent to 46 percent of HR-positive, HER2-negative breast cancers and are associated with a poor prognosis.

The SOLAR-1 results are generally viewed in the field as a confirmation that inhibiting mutant PI3Kα leads to anti-tumor activity but that these drugs can also inhibit wild-type PI3Kα, causing hyperglycemia. A study published in Cell Metabolism in 2019 casts doubt on that theory, however.

The researchers in that study investigated the roles of the various PI3K isoforms in insulin signaling using isoform-specific PI3K inhibitors in cell lines and in genetically modified mice and concluded that insulin signaling depends on redundant PI3Kα and PI3Kβ activity. That raises the possibility, according to the researchers, that the hyperglycemia seen in patients treated with mutant PI3Kα inhibitors such as Piqray could be caused by their off-target effects on PI3Kβ, which otherwise should be able to compensate for inhibition of PI3Kα by maintaining normal insulin signaling in the cell.

"Whereas the current model for insulin signaling cannot explain why [Piqray] displays a higher hyperglycemic threshold than pan-PI3K inhibitors, our improved model for insulin signaling predicts that [Piqray] should cause hyperglycemia only at doses where its specificity for PI3Kα is lost," the study authors wrote.

Managing hyperglycemia

The theory of why hyperglycemia occurs that has been advanced by studies supporting Piqray's approval has taken hold among drugmakers that are developing mutant PI3Kα inhibitors and seeing the same class effect.

For example, Roche subsidiary Genentech reported results from the Phase III INAVO120 study this month showing that its mutant PI3Kα inhibitor inavolisib combined with Pfizer's CDK 4/6 inhibitor Ibrance (palbociclib) and fulvestrant reduced the risk of median progression-free survival by 57 percent compared to placebo-Ibrance-fulvestrant. There were also positive trends in the study that the addition of inavolisib improves overall survival, objective response rate, duration of response, and the clinical benefit rate. Still, in the study, 5.6 percent of patients on the inavolisib combination experienced grade 3 or higher hyperglycemia, compared to none of the patients in the placebo arm.

In a long-term safety analysis from a Phase I/Ib study of inavolisib with fulvestrant in patients with PIK3CA-mutated, HR-positive, HER2-negative metastatic breast cancer, researchers at the American Society of Clinical Oncology's annual meeting last year reported that 125 out of 193 patients, or 65 percent, experienced hyperglycemia, and 42 patients, or 22 percent, had grade 3 or higher hyperglycemia.

According to a Roche spokesperson, the company believes hyperglycemia occurs when inavolisib inhibits wild-type PI3Kα, and that its incidence may depend on molecule-specific properties, including the potency of PI3Kα inhibition and selectivity of wild-type versus mutant PI3Kα inhibition.

With inavolisib, however, Genentech is taking cues from Novartis' strategies for managing Piqray's hyperglycemia side effects. For example, Novartis advises doctors to test fasting plasma glucose and HbA1c levels and optimize patients' blood glucose before initiating treatment. During treatment, the company advises physicians to routinely monitor patients' blood sugar. If hyperglycemia occurs, depending on its severity, patients may need to have their Piqray doses interrupted, reduced, or discontinued.

Doctors can also use metformin to control blood sugar effects, according to Piqray's label.

In results from the METALLICA study, presented at the 2022 San Antonio Breast Cancer Symposium, researchers found that the use of prophylactic metformin reduced all-grade hyperglycemia by almost 80 percent in patients taking Piqray.

With Genentech's inavolisib, a Roche spokesperson said that "hyperglycemia … has been managed with oral anti-hyperglycemic medications and inavolisib dose modifications per management guidance provided in study protocols."

While Novartis and Genentech are confident that hyperglycemia seen with Piqray and inavolisib can be successfully managed, the risk of this and other toxicities associated with PI3K inhibitors have hampered their development.

Inavolisib is not Genentech's first outing in the PI3Kα space, for example. In 2018, the company discontinued development of taselisib, a dual PI3Kα/PI3Kδ inhibitor with enhanced specificity for mutant PI3Kα, following disappointing results from a Phase III trial. In that trial, patients with estrogen receptor (ER)-positive, HER2-negative, PIK3CA-mutated breast cancer had a 30 percent lower chance of their cancer worsening with taselisib and fulvestrant than on fulvestrant alone. However, 50 percent of patients on taselisib-fulvestrant had grade 3 or higher toxicities, compared to 16 percent of patients in the comparator arm, and 40.4 percent of patients on taselisib-fulvestrant had hyperglycemia, compared with 9.4 percent on fulvestrant alone.

Hyperglycemia-free mutant PI3Kα inhibitors?

Smaller biotech companies see an opportunity here to develop a mutant PI3Kα inhibitor with minimal or no hyperglycemic side effects.

Relay is advancing two pan-mutant PI3Kα inhibitors, RLY-2608 and RLY-5836. Both compounds inhibit the H1047X, E542X, and E545X mutants of PI3Kα. Relay also has identified mechanisms for targeting PI3Kα inhibitors that are specific to a single mutation, such as H1047X. "We have an active drug discovery program against H1047, but we prioritized going with a pan-mutant selective approach because it allowed us to fully address the entirety of the patient population with mutant PI3Kα-driven tumors," said Don Bergstrom, Relay's head of R&D.

According to Bergstrom, Relay's decision to develop mutant-selective PI3Kα inhibitors was based on clinical trial data from non-mutant selective PI3Kα inhibitors. "The entire industry has been working for 20 years toward mutant selectivity," Bergstrom said.

Relay's leadership is aware of the 2019 Cell Metabolism paper that put forth the alternative theory that hyperglycemia is caused by off-target effects of mutant PI3Kα inhibitors on PI3Kβ, but Bergstrom isn't an acolyte of this hypothesis. First, he said the study was a preclinical analysis conducted at a time when the field already had clinical data on the incidence of hyperglycemia from PI3Kα-specific inhibitor studies. "Additionally, this paper focuses only on PI3Kα- and β-specific knockouts, which could have a different impact than transitory inhibition," he said, further pointing out that the analysis focused on drug uptake in liver tissue only and did not address uptake in skeletal muscles, which could function differently.

Bergstrom instead highlighted a study published in 2021 in the International Journal of Molecular Sciences, which, in his view, lends further support to the dominant theory that inhibition of wild-type PI3Kα is the main culprit causing hyperglycemia. In that study, researchers explored the specific functions of PI3Kα and PI3Kβ in mouse adipocytes using isoform-specific inhibitors and concluded that α is the principal isoform that is activated downstream of insulin receptor activation. According to Bergstrom, that points to PI3Kα as the primary driver of glucose uptake, and "therefore its inhibition would reduce glucose uptake and result in hyperglycemia."

The authors of that study, however, also noted that PI3Kβ can substitute for PI3Kα in hepatocytes from liver-specific PI3Kα-deleted mice, suggesting that PI3Kα and PI3Kβ are redundant in the insulin signaling pathway in the liver.

But for drugmakers that believe that inhibition of wild-type PI3Kα is the main problem, developing inhibitors that can selectively home in on mutant PI3Kα but spare the wild-type form presents a promising strategy for avoiding hyperglycemia. In preclinical studies, Relay's RLY-2608 has demonstrated selectivity for mutant PI3Kα over the wild-type form and other isoforms of PI3K, and in patient-derived xenograft models, combining RLY-2608 with standard-of-care therapies has caused ER-positive, HER2-negative breast cancers to regress.

Encouraged, Relay in December 2021 began testing RLY-2608 in a Phase I trial, dubbed ReDiscover, both as a single agent in patients with unresectable or metastatic PIK3CA-mutated solid tumors and in combination with fulvestrant in patients with HR-positive, HER2-negative, PIK3CA-mutated advanced breast cancer. In April, the company also began a first-in-human trial of RLY-5836, another mutant-selective, oral PI3Kα inhibitor, as a single agent in patients with advanced PIK3CA-mutant solid tumors and in combination with targeted and endocrine therapies in patients with PIK3CA-mutant, HR-positive, HER2-negative advanced breast cancer.

In preliminary results from the ReDiscover trial reported at the American Association for Cancer Research's annual meeting in April, among 43 patients who had been enrolled thus far, none had yet experienced a dose-limiting toxicity on RLY-2608, and while 17 percent had low-grade hyperglycemia, none experienced grade 3 or higher hyperglycemia. Thirty-one percent of patients needed dose interruptions, and 2 percent needed dose reductions.

In terms of efficacy, nine out of 16 breast cancer patients, or 56 percent, had radiographic tumor reductions on RLY-2608; 12 had stable disease; and one had an unconfirmed partial response. Eleven patients were still responding to treatment at the time of the data cutoff.

In one breast cancer patient who had received RLY-2608 monotherapy, circulating tumor DNA analysis showed a more than 90 percent reduction in variant allele frequency of PIK3CA and ESR1 mutations after four weeks. After eight weeks, the patient's scans showed a 36 percent tumor reduction. This patient had no treatment-related adverse effects.

"We are very happy with the data that we saw because it was very consistent with the preclinical profile of [RLY-2608]," Bergstrom said.

While Relay has interpreted these initial results as signs that the RLY-2608 program is on the right track, in a report in July, research analysts Sam Slutsky and Anshul Dhankher with the investment bank LifeSci Capital called RLY-2608's performance in ReDiscover "sub-par." They noted that 81 percent of patients given Piqray with fulvestrant in a Phase I study experienced some tumor shrinkage.

Slutsky and Dhankher suggested that the relatively weaker performance of RLY-2608 may be due to possible issues with potency, pharmacokinetics, and the early stage of the study, but they also questioned whether wild-type PI3Kα may still be driving tumor growth since RLY-2608 is designed to specifically target mutant PI3Kα. "Because many [tumor] cells will have heterozygous mutations, it is possible that inhibition of the mutant form will cause a compensatory upregulation of [wild-type PI3Kα] and ongoing proliferation," they wrote.

Despite the limited efficacy seen so far in ReDiscover, in Slutsky and Dhankher's view, Relay's data set the stage for what the field might expect from other mutant-selective PI3Kα inhibitors in development when it comes to the incidence of hyperglycemia and other toxicities. "With just [5- to 10-fold] selectivity for H1047R over wild-type PI3Kα, Relay saw only low-grade hyperglycemia and very little rash or diarrhea, setting the bar for what future programs could expect based on their molecular properties," they wrote.

Scorpion Therapeutics CEO Axel Hoos said his firm also sees value in targeting mutant PI3Kα and is developing STX-478, a mutant-selective PI3Kα inhibitor that is entering clinical testing after having completed preclinical studies. "We chose [PI3Kα] because it's an underserved pathway with no good drugs available yet," said Hoos, adding that it was important at Scorpion to create a compound that is "very selective for the mutations that only exist in the cancer cell."

A mutant-selective PI3Kα inhibitor, according to Hoos, would also have to have a wider therapeutic window because then it could be given at sufficient doses that inhibit the PI3Kα pathway without inducing hyperglycemia, based on the company's preclinical studies. "[Piqray] cannot be dosed to the degree that it hits the pathway in full, at least not in humans," Hoos said. "You never get full pathway inhibition because toxicity prevents that."

Novartis noted that preclinical testing and clinical testing yield distinct, and not necessarily equivalent, insights, particularly in the realm of tolerable dosing. "This does not in any way diminish the pathway inhibition demonstrated in the clinical trial setting [of Piqray]. Novartis is not aware of any credible reference that proves any drug's ability to replicate preclinical findings in the clinical setting," said a Novartis spokesperson.

As for Relay's RLY-2608, Hoos said that while the agent has "selectivity," Relay hasn't been able to fully address metabolic dysregulation. This is evidenced, according to Hoos, by hyperinsulinemia in preclinical models.

In comparison, Hoos said Scorpion's STX-478 has favorable druglike qualities, as demonstrated in preclinical in vivo studies, including low molecular weight, a "clean" pharmacokinetic profile, and a lower dosing range. Relay, he noted, has tested 800 mg doses of RLY-2608 in clinical studies while, based on preclinical data, Scorpion anticipates using doses as low as 50 mg in its trials.

Relay's Bergstrom pointed out that their compound differs from Scorpion's in that RLY-2608 is a pan-mutant inhibitor. "It's interesting to note that [Scorpion] appears to be focusing on the [H1047-mutant] population for initial development of the asset," he said. "One of the advantages we see for [RLY-2608] that's been supported by clinical data that we've generated so far is that [RLY-2608] does appear to have activity independent of what the underlying PIK3CA mutation is."

Scorpion's STX-478 zeroes in on different PIK3CA H1047 mutations, including the most common one, H1047R, which represents 40 percent to 50 percent of all PI3Kα mutants. In preclinical studies, STX-478 had activity in patient-derived xenograft models of ER-positive breast cancer with PI3Kα mutations and was well tolerated. The drug did not cause hyperglycemia in rodents or in dogs. Another difference, Hoos noted, is that STX-478 penetrates the brain, giving it the potential to treat central nervous system metastases.

In May, Scorpion began a Phase I/II dose-escalation and -expansion trial of STX-478. Initially, investigators are enrolling patients with HR-positive, HER2-negative breast cancer. Once a dose is established, researchers will begin enrolling other patient cohorts, including those with PIK3CA-mutated gynecologic, head and neck, and other solid cancers. In the trial, investigators will evaluate STX-478 as a monotherapy and in combination with fulvestrant.

Scorpion remains confident in its strategy of targeting mutant PI3Kα and sparing wild-type PI3Kα. "Preclinical and clinical data indicate that hyperglycemia is an on-target effect of inhibiting wild-type PI3Kα, and we expect this as a dose-limiting toxicity of any drug that inhibits wild-type PI3Kα," Hoos said.

A no-regrets selectivity profile

In contrast to companies like Scorpion, Relay, and Novartis, which are aiming to address the largest possible patient population with a mutant PI3Kα inhibitor, Loxo Oncology, an Eli Lilly company, is betting that its drug, LOXO-783, which narrowly targets H1047R mutations, can completely eliminate the hyperglycemia effects of the drug class. The H1047R mutation comprises half of the PI3Kα mutations found in breast cancer and shows up in about 15 percent of patients across the different breast cancer subtypes.

Although targeting just one mutation reduces the addressable patient population for a PI3Kα inhibitor by about half, Eli Lilly Chief Medical Officer David Hyman said "our medicine is exquisitely targeted to this one mutation with a no regrets profile from our perspective on its selectivity for that one mutation versus the normal protein."

In comparison, Hyman said Relay's and Scorpion's programs are similar in that their compounds are designed to be active against a broader range of mutations in PI3Kα. While these drugs may potentially address a larger patient population, in terms of selectivity, they are not as strong as Loxo's compound, in his view.

Given the design of RLY-2608, for example, "you might expect that it would cause some hyperglycemia," Hyman said. "In fact, that's what the early clinical data that we saw showed," he added, referring to the low-grade hyperglycemia seen in 17 percent of patients in ReDiscover.

Hyman attributes the high selectivity of LOXO-783 to a different binding mechanism than other PI3Kα inhibitors. Most inhibitors bind to the catalytic site of the protein but LOXO-783 is an allosteric inhibitor, meaning that it binds in a pocket distant from the active site. Instead of physically blocking the active site, it triggers a conformational change that renders the protein unable to carry out its signaling activity. Like Scorpion's drug, LOXO-783 also has the ability to penetrate the brain, giving it potential to treat brain metastases.

In preclinical studies, Hyman said LOXO-783 has activity that matches or exceeds that of Piqray, and Loxo researchers saw synergistic effects when they combined LOXO-783 with endocrine or other common breast cancer therapies. In animal studies, there were no signs of hyperglycemia.

"There is a compromise to be made between the breadth of mutations that you can address and the selectivity with which you can address them," Hyman said. "That's the design decision we made, and we'll see how that decision plays out vis-à-vis others working in this space."

In an open-label, Phase I clinical trial, dubbed PIKASSO-1, Loxo is testing LOXO-783 as a monotherapy and in combination with other therapies in around 400 advanced breast cancer and other tumors with a PIK3CA H1047R mutation. In the Phase Ia dose-escalation part of the trial, patients will receive LOXO-783 alone as investigators track toxicities and identify a maximum-tolerated dose and a recommended Phase II dose. In the Phase Ib portion, patients will receive combinations of LOXO-783 with fulvestrant, other aromatase inhibitors, and Eli Lilly's CDK4/6 inhibitor Verzenio (abemaciclib). The investigators are tracking efficacy outcomes, including overall response rate, progression-free survival, and overall survival.

Loxo is in no rush to release data from this trial. "We are in the fortunate position of not having to have data disclosures driven by the need to raise capital for the program," Hyman said. "We tend to disclose the data when we have a story to tell that is actually interpretable for the medicine."

A contrarian approach

While Scorpion, Relay, Loxo, and others have focused on targeting mutant PI3Kα as the solution to avoiding or reducing hyperglycemia associated with PI3Kα inhibitors, Totus is operating on the alternate theory that hyperglycemia is not due to inhibition of wild-type PI3Kα, but due to off-target inhibition of PI3Kβ.

Neil Dhawan, Totus CEO and head of R&D, said that reversible inhibitors of mutant PI3Kα in development at other companies are incapable of completely suppressing PI3Kα signaling to induce tumor cell death and regression. To address that shortcoming, Totus has developed a covalent inhibitor that irreversibly binds to wild-type and mutant PI3Kα, which the company believes will lead to better efficacy and safety due to the specificity of the drug for the α isoform over β.

Dhawan pointed out that although Totus' competitors claim that targeting wild-type PI3Kα will lead to toxicity, and particularly to hyperglycemia, recent studies have shown "if you don't inhibit wild-type PI3Kα as well as mutant [PI3kα], you cannot achieve full efficacy, tumor cell death, and regression." This is the case, Dhawan reasoned, because once mutant PI3Kα is inhibited, wild-type PI3Kα signaling becomes activated, leading to continuous signaling and continuous tumor cell proliferation.

"The field has been worried that if you inhibit wild-type PI3Kα, it will lead to chronically high glucose levels, but it turns out knockouts of PI3Kα do not lead to sustained high [glucose levels]," Dhawan said, citing an animal study in which knocking out PI3Kα in mice did not induce hyperglycemia.

In contrast, Totus' strategy is informed by newer research suggesting that PI3Kβ compensates for loss of PI3Kα in insulin signaling. "You have to knock out [PI3Kα] and [PI3Kβ] to get sustained hyperglycemia," he said.

According to Dhawan, Totus has been able to achieve more than 95 percent inhibition of PI3Kα with its drug, TOS-358. This is a feat that, he said, "has not been possible with the [other] reversible, mutant-selective approaches in this space," and the company has not seen major toxicities with TOS-358 as a result.

In April, Totus launched a Phase I trial of TOS-358 in patients with PIK3CA-mutated cancer. The company is testing safety, tolerability, and pharmacokinetics of the drug in 241 patients with solid tumors, particularly colorectal, gastric, non-small cell lung, HER2-negative breast, urothelial, squamous cell head and neck, and gynecologic cancers. The company hopes to determine a maximum-tolerated dose, a recommended Phase II dose, and safety and tolerability at the recommended Phase II dose.

Dhawan said Totus is hoping to share data from the trial in the first half of 2024.

So far, Totus' approach has been an outlier in the PI3Kα inhibitor space. Despite the company's argument that hyperglycemia could be the result of off-target effects on PI3Kβ with Piqray, in their report to investors in July, LifeSci Capital research analysts Slutsky and Dhankher pointed out that Genentech's inavolisib is 3,000-fold more selective for the α isoform over β. And yet, in a Phase I trial, there was a 22 percent rate of grade 3 hyperglycemia among patients on a low dose of inavolisib, casting some doubt on this alternative hypothesis.

"Even without [pharmacokinetic] data, it's much less likely that inavolisib is also inhibiting PI3Kβ, yet the drug is still causing high levels of grade 3 hyperglycemia," they wrote. "By this logic, it is unlikely that off-target PI3Kβ inhibition is driving hyperglycemia."

Dhawan maintains that Totus' approach is data-driven and that the company has taken into account the totality of the literature. "We're very excited for this to play out in 2024," he said.

Not just Totus, but Relay is also expecting further Phase I data on its agent next year, while Scorpion is awaiting a first-time Phase I readout and Genentech plans to report more detailed Phase III results. As such, 2024 might be the year that the field finally gets answers to longstanding questions about PI3Kα's role in cancer and hyperglycemia, which could herald advancement of one or more practice-changing therapies for breast and other cancers.