Improving the Armamentarium of PI3K Inhibitors with Isoform-Selective Agents: A New Light in the Darkness

That the PI3K pathway plays a relevant role in cancer, as one of the most frequently activated pathogenic signaling routes in human cancers, has been known for decades (1). A large body of literature has been compiled in this field, and the PanCancer analysis of The Cancer Genome Atlas just confirmed that a multitude of alterations are frequently present in many tumor types (2), including mutations and/or amplification of PIK3CA, AKT1-3, and PTEN or INPP4B. These alterations are known to induce a transformed phenotype and drug resistance and consequently pose PI3K pathway components as attractive targets for anticancer drug development. Structural studies of PI3K enzymes and medicinal chemistry efforts enabled the development of sufficiently specific drugs with favorable drug properties to suggest that their development could become another archetype in precision medicine (1). Drug development searching for inhibiting nodes of the PI3K pathway blossomed and brought an onrush of drug companies seeking to regulate this pleiotropically altered pathway. In this urgency to hit the market, data indicating the complexity of PI3K signaling, redundancies, and cross-talk with other pathways capitulated to the need for oversimplification of conventional clinical trials (which can answer only a limited number of simple questions). Evidence challenging the paradigm of PIK3CA mutations behaving as a classic oncogene addiction scenario was neglected, while results of the next blockbuster were to be anticipated. When only one regulatory approval of one such agent, idelalisib (Gilead Sciences) was achieved (in several hematologic malignancies), the gold fever vanished with a therapeutic ice bath. Following the explosion in drug development, an implosion: Pharma companies moved on, in pursuit of the next “gold rush” including the one promised by immuno-oncology, and closed many PI3K inhibitor development programs (Table 1). In breast cancer, for example, one of the most promising areas for the development of PI3K inhibitors, positive results with mTOR or PI3K inhibitors in hormone-sensitive disease seem to now be eclipsed by the shining cyclin-dependent kinase inhibitors, which have unquestionable antitumor efficacy, have side effects manageable in a busy clinic, and require no intricate predictive markers.

Looking back to 2012, when more than 30 drugs targeting different nodes of the PI3K pathway were in clinical development, if we now revisit the open questions of that time (many phase I trials of these drugs already were completed by then), many of them still remain unsolved today. To name but a few:

• (i) Which of the different agents, whether mTOR, AKT, pan-PI3K, or isoform-specific PI3K inhibitors, provide the greatest therapeutic index? Is there a specific genetic context that prognosticates superior activity for each class?

• (ii) What magnitude of signaling inhibition is required to produce biological (apoptosis) and clinical effects? If chronic inhibition at that level is not sustainable due to side effects (either off-target effects or insufficient control of on-target toxicities), is intermittent dosing sufficient to achieve the desired effect?

• (iii) If there are redundant components of the pathway and compensatory intrapathway or extrapathway mechanisms of resistance, which combinations would make sense in cancer patients (rather than in artificial in vitro/in vivo models)?

• (iv) Which patients, selected on the basis of which biomarkers, are more likely to benefit from these agents?

Nowadays, though, some progress has actually shed light in this field, albeit away from the fuss of harsh industrial competition, and new insights are providing a deeper understanding of the pathway and its pharmacology. Such discovery may ultimately deliver on the full potential of PI3K inhibitors. On one hand, pharmacologic research is following two natural paths to overcome some of the toxicity observed with pan-PI3K inhibitors (and these are not mutually exclusive solutions): to explore new scheduling alternatives that may be less toxic and to develop isotype-specific PI3K inhibitors with different, maybe better, specificity and safety profiles, including isoform-specific, or “isoform-balanced” (idelalisib - PI3Kδ, alpelisib - PI3Kα, copanlisib - PI3Kα/δ, taselisib - PI3Kα/δ/γ…). On the other, the molecular biology of the pathway is being explored in novel ways. Translational research using new research paradigms, models, methods, and tools such as patient-derived xenografts, cell-free DNA, and warm autopsies has certainly helped in these efforts. Through these new approaches in molecular biology, we are achieving (i) a better understanding of the role of PI3K and the different isoforms, their signaling and redundant wiring as well as the potency needed for antitumor effect; (ii) knowledge of how PI3K isoforms and family components contribute to other hallmarks of cancer, such as metabolism, inflammation, angiogenesis, and immunosurveillance; and (iii) discerning the role of heterogeneity and clone evolution, and how they may drive secondary resistance (3). Let us take breast cancer, for instance, and delineate some of the advancements driven by translational research:

• Our understanding of the cross-talk between estrogen receptor (ER) and PI3K has certainly improved. Recognizing that PI3K pathway alterations in hormone-sensitive tumors act as resistance mechanisms to antiestrogen therapies, we now know that both pathways are interdependent and dual inhibition is needed in any case, even in PIK3CA wild-type cases (4). This could be achieved with different schedules of pharmacologic inhibition that, although all efficacious, may have different pharmacodynamic effects and therapeutic index (5).

• Clonal evolution in ER-positive breast cancer and emergence of PIK3CA mutations seems important as a predictive biomarker: In a phase III study of buparlisib (a pan-PI3K inhibitor) in ER-positive, HER2-negative advanced breast cancer, patients with PIK3CA-mutant status detected in circulating tumor DNA performed poorly on fulvestrant alone, but obtained benefit from the addition of the PI3K inhibitor to the regimen (6).

• On-treatment biopsies and warm autopsies upon progression can also help in depicting secondary mechanisms of resistance in this field. Of the multiple mechanisms of primary resistance described in in vitro/in vivo models, in clinical samples secondary mutations in PTEN (7) and/or increased PDK1–SGK1 signaling (8) have been observed, both sustaining residual downstream mTORC1 activity and indicating a need for potent downstream inhibition.

• Second-generation drugs, such as the PI3Kα-specific inhibitors alpelisib and serabelisib, show better safety profiles and probably an increased antitumor efficacy compared with what was previously seen in ER-positive breast cancer (patients with PIK3CA mutation treated with alpelisib had a disease control rates of 53% in monotherapy and 80% in combination with antihormonal therapy; ref. 9).

In this new context, Juric and colleagues (10) present the results of taselisib, a “p110β-sparing” PI3K inhibitor. In a way, it leverages some of this knowledge and represents one of the members of the third wave of drug development of PI3K inhibitors (if rapalogs represent the first and PI3K/mTORC/AKT inhibitors embody the second). Some results may not be especially novel, such as their safety profile or pharmacodynamic markers. The observed side effects are well known in this drug family and probably explained as a class effect (colitis expected from PI3Kγ inhibition, hyperglycemia from PI3Kα inhibition). Results from biomarkers such as surrogate tissues, metabolic markers, and FDG-PET scan confirm prior observations with other PI3K inhibitors. Neither of these (safety observations or pharmacodynamic markers) greatly illuminates the role of the different isoforms (one wonders whether at the recommended doses, p110β is indeed “spared”). The antitumor effects are however very intriguing. In hormone-sensitive breast cancer, and especially in PIK3CA-mutant tumors, both taselisib and alpelisib seem to be achieving significant antitumor effect as monotherapy, and even more so in combination with antihormone therapy. These “new kids on the block” are showing significant promise, and this is well exemplified here by the trial of Juric and colleagues. Considering the new body of emerging evidence, a second wave of clinical research and drug development with PI3K inhibitors is to be expected. It is time for science again, and to let translational research do the talking.

J. Rodon reports receiving commercial research grants from Bayer and Novartis and is a consultant/advisory board member for Ability, Novartis, Orion, and Peptomyc. J. Tabernero is a consultant/advisory board member for Genentech and Novartis.

The University of Texas MD Anderson Cancer Center is supported by the NIH Cancer Center support grant CA016672.