Molecular switches in cell death: the interplay of apoptosis and pyroptosis in cancer

Abstract

Programmed cell death in cancer encompasses apoptosis, pyroptosis, necroptosis, ferroptosis, autophagy-dependent cell death and other interconnected programs. Within this broader landscape, apoptosis and pyroptosis are especially important because the same upstream anticancer pressure can generate different biological outputs: non-lytic apoptotic clearance, gasdermin-mediated lytic inflammation, or mixed death phenotypes. Rather than cataloguing these pathways as separate descriptions, this review integrates them as a cancer-specific switching network. It organizes the evidence around a sequence of determinants: the initiating tumor stress, the active caspase repertoire, the availability and cleavage site of gasdermin substrates, the cellular compartment in which death occurs, and the resulting immune or inflammatory output. The review discusses caspase-3-mediated cleavage of gasdermin E (GSDME), caspase-8-mediated regulation or cleavage of GSDMD and GSDMC, canonical and noncanonical inflammasome signaling, cFLIP, granzyme-mediated gasdermin cleavage, and autophagy-dependent removal of inflammasome stimuli. It distinguishes mechanisms shown in infection or myeloid-cell models from evidence demonstrated in tumor cells or tumor microenvironment models. This framework highlights when pyroptosis may overcome apoptotic resistance and enhance antitumor immunity, and when excessive or poorly localized pyroptosis may promote tissue injury, immune suppression or chronic tumor-supportive inflammation. A mechanistic, cancer-specific interpretation of apoptosis-pyroptosis crosstalk may support biomarker-guided therapeutic strategies that exploit immunogenic tumor-cell death while limiting inflammatory toxicity.

Introduction

Cell death is not merely a biological endpoint, but a regulated physiological process essential for development, tissue remodeling and organismal homeostasis [1, 2]. In cancer, however, the same death programs become therapeutically consequential because malignant cells can evade, delay or redirect them under drug and immune pressure. Programmed cell death (PCD) now includes a broad spectrum of modalities, including apoptosis, pyroptosis, necroptosis, ferroptosis, autophagy-dependent cell death, anoikis, mitotic catastrophe and paraptosis [3–6]. These pathways should not be viewed as isolated routes. Specific shared modules include death-receptor FADD/RIPK1/caspase-8 signaling, which can support apoptosis, necroptosis control or inflammatory death; mitochondrial caspase-9-caspase-3 signaling, which can switch from apoptosis to GSDME-dependent pyroptosis; NLRP3/AIM2-ASC-caspase-1-GSDMD signaling, which processes IL-1β/IL-18 and cleaves GSDMD; apoptotic-caspase regulation of gasdermins; and autophagy-mediated removal of damaged mitochondria, cytosolic microbial stimuli, damaged ER fragments and protein aggregates that would otherwise prime inflammasomes [7–9]. These shared modules are especially important in cancer, where genotype, gasdermin silencing, hypoxia, immune infiltration and therapy-induced stress determine whether cell death is silent or inflammatory. Accordingly, this review is organized around the mechanistic decision that connects apoptosis to pyroptosis in cancer: which upstream stress activates which caspase, which gasdermin substrate is available in which cell type, and whether the final output promotes tumor-cell elimination, antitumor immunity, immune suppression or inflammatory toxicity.

Apoptosis

First conceptualized in 1972 [10], apoptosis is a regulated, usually non-lytic form of cell death characterized by cell shrinkage, phosphatidylserine externalization, membrane blebbing, chromatin condensation and fragmentation into apoptotic bodies [11–13]. These bodies are rapidly cleared by phagocytes and therefore usually do not generate the same inflammatory burst as gasdermin pore formation. For cancer, the importance of apoptosis is not only that malignant cells evade it through p53 loss, Bcl-2 family imbalance, death-receptor defects or IAP activity; it is also that the apoptotic machinery supplies the caspase activity that can be redirected toward pyroptosis when gasdermin substrates are present. The extrinsic, intrinsic and ER-stress pathways are therefore discussed below as entry routes into a shared caspase decision module rather than as isolated textbook pathways.

Extrinsic pathway

The extrinsic apoptosis pathway, commonly referred to as the death receptor pathway, is triggered by the ligation of death ligands to their corresponding transmembrane receptors [14]. Prominent among these are the type 1 tumor necrosis factor (TNF) receptor (TNFR1) and the Fas receptor (CD95), which are activated by TNF and Fas ligand (FasL), respectively (Fig. 1) [15]. Upon ligand binding, the intracellular death domains of these receptors recruit specific adaptor proteins, notably the TNF receptor-associated death domain (TRADD) and the Fas-associated death domain (FADD) [16]. This receptor-adaptor interaction provides a structural scaffold that culminates in the assembly of the death-inducing signaling complex (DISC) [17]. The DISC subsequently orchestrates the recruitment and activation of the initiator caspase, pro-caspase 8, which then directly activates the executioner caspase 3 to drive the apoptotic program [18, 19]. In cancer, however, death-receptor signaling should be interpreted as a branching node rather than a fixed apoptotic route. DISC assembly, cFLIP abundance, caspase-8 phosphorylation, IAP activity and gasdermin expression can determine whether TNF/Fas signals produce apoptosis, GSDMC- or GSDME-linked pyroptosis, or alternative inflammatory death programs. This is why death-ligand-based therapy and immune-cell killing need to be evaluated together with the tumor-cell caspase and gasdermin state.

Fig. 1.

Classical apoptosis pathways. Apoptosis can be initiated via three primary convergent routes: the extrinsic (death receptor) pathway, the intrinsic (mitochondrial) pathway, and the endoplasmic reticulum (ER) stress pathway. The three pathways converge on caspase 3, which usually produces apoptotic morphology and silent clearance but can support inflammatory output in selected contexts

Intrinsic pathway

The intrinsic, or mitochondrial, pathway is initiated by intracellular stress signals such as DNA damage, hypoxia, cytosolic calcium overload, and elevated oxidative stress (Fig. 1) [20, 21]. Its central event is mitochondrial outer membrane permeabilization (MOMP), which releases cytochrome c, Smac/DIABLO, Omi/HtrA2 and other pro-apoptotic factors into the cytosol [22–24]. MOMP is governed by the balance between pro-apoptotic effectors and anti-apoptotic Bcl-2 family proteins, and this balance is frequently distorted in cancer by oncogenic survival signaling or therapeutic selection pressure [25, 26]. Once released, cytochrome c binds Apaf-1 and pro-caspase 9 to form the apoptosome, while Smac/DIABLO and Omi/HtrA2 neutralize IAP-mediated caspase inhibition; together these events activate caspase 3 [27, 28]. The cancer-specific implication is that mitochondrial apoptosis is the main route by which many chemotherapies and DNA-damaging treatments activate caspase 3. If GSDME is absent or epigenetically silenced, this produces relatively silent apoptosis; if GSDME is expressed, the same mitochondrial cascade can become lytic, release DAMPs and reshape immune recognition. Thus, the therapeutic meaning of mitochondrial apoptosis depends on the downstream gasdermin state.

Endoplasmic reticulum pathway

Serving as a third mechanism, the endoplasmic reticulum (ER) pathway is triggered by severe ER stress (Fig. 1). Insults such as hypoxia, oxidative stress, or glucose deprivation disrupt ER homeostasis, leading to the accumulation of unfolded or misfolded proteins [29, 30]. Under these conditions, TNF receptor-associated factor 2 (TRAF2) mediates the activation of caspase 12, which initiates a cascade culminating in caspase 3 activation and apoptosis [31, 32]. Furthermore, caspase 12 can be processed and activated by m-calpain in response to elevated intracellular Ca²⁺ levels. Once active, caspase 12 is capable of directly cleaving caspase 9, thereby driving apoptosis in a predominantly mitochondria-independent manner. Given its critical role in sensing localized cellular stress and amplifying apoptotic signaling, the ER pathway has increasingly become a focal point of cell death research [33]. In tumors, ER stress should also be connected to the apoptosis-pyroptosis interface because hypoxia, nutrient deprivation and therapy-induced proteotoxic stress can increase mitochondrial damage, ROS production and DAMP release. These changes may not directly prove tumor-cell pyroptosis, but they can prime inflammasome activation in myeloid cells and influence whether caspase-3 activation is immunologically silent or inflammatory.

Common pathway

The execution phase of apoptosis is fundamentally characterized by the activation of effector caspases (Fig. 1). While the extrinsic and intrinsic pathways use different initiator caspases, caspase 8 and caspase 9, respectively, both signaling cascades commonly converge on caspase 3 [34]. As the primary executioner, activated caspase 3 leaves substrates involved in DNA repair, cytoskeletal integrity, survival signaling and cell-cycle control, producing the classical morphology of apoptosis [35–38]. In cancer, this pathway is frequently restrained by p53 dysfunction, increased Bcl-2 family survival signaling, IAP activity or impaired death-receptor signaling [14, 39]. Importantly, apoptotic signaling can also connect to inflammatory pathways in a context-dependent manner. In caspase-1-deficient macrophages infected with Francisella, for example, the AIM2/ASC platform can support caspase 8-dependent apoptosis rather than canonical caspase-1-mediated pyroptosis [40]. This infection-model observation illustrates pathway plasticity, but it should not be directly extrapolated to tumor cells without cancer-specific validation.

Pyroptosis

Pyroptosis is a lytic and inflammatory form of regulated cell death characterized by gasdermin pore formation, cellular swelling, plasma-membrane rupture and the release of cytokines and danger signals [40–42]. In contrast to apoptosis, which is usually resolved by efferocytosis without strong inflammation, pyroptosis can expose tumor antigens, ATP, HMGB1, mitochondrial DNA and mature IL-1β/IL-18 to the extracellular space. These outputs can recruit and activate immune cells, but they can also amplify chronic inflammatory loops when activation is sustained [40]. Therefore, the term pyroptosis should be used with mechanistic precision: in current usage it is primarily a gasdermin-mediated process, whereas gasdermin-independent inflammasome-associated lysis or apoptosis-like death should be described separately unless gasdermin pore formation is demonstrated.

Mechanistically, pyroptotic signaling is commonly divided into canonical and noncanonical pathways (Fig. 2). In the canonical pathway, sensors such as NLRP3, NLRP1, NLRC4 or AIM2 assemble inflammasomes that recruit ASC and activate caspase 1. Active caspase 1 cleaves GSDMD and processes pro-IL-1β and pro-IL-18, thereby coupling pore formation to cytokine maturation. In the noncanonical pathway, human caspases 4 and 5 or mouse caspases 11 directly sense cytosolic lipopolysaccharide and cleave GSDMD, which can secondarily activate NLRP3 through potassium efflux and membrane damage [43]. In cancer, canonical inflammasome activity is often most prominent in myeloid cells of the tumor microenvironment, whereas tumor-cell pyroptosis depends strongly on gasdermin expression, caspase availability and the specific therapeutic stimulus. This distinction is important because observations from macrophages or infection models do not always predict the behavior of epithelial tumor cells.

Fig. 2.

Canonical and noncanonical pyroptosis. Canonical Pyroptosis Pathway: Triggered by PAMPs/DAMPs. Noncanonical Pyroptosis Pathway: Initiated by the direct binding of intracellular LPS to the CARD domain of pro-caspase-11 (in mice) or pro-caspase-4/5 (in humans). Inflammasome-caspase 1 and cytosolic LPS-caspase 4/5/11 pathways cleave GSDMD, forming pores and releasing IL-1β, IL-18 and DAMPs

Inflammasomes function as intracellular sentinels that monitor pathogen-associated molecular patterns (PAMPs) and damage-associated molecular patterns (DAMPs), including microbial ligands, extracellular ATP, uric acid crystals, oxidized mitochondrial components and disrupted organelles [44, 45]. Structurally, a typical inflammasome contains a cytosolic sensor, the adaptor ASC and pro-caspase 1 [46]. In a purely descriptive review, the emphasis would be on listing sensors such as NLRP3, NLRP1, NLRC4, AIM2 and IFI16 [47]. For cancer interpretation, however, the more important issue is where and why these sensors are activated. Acute inflammasome activation in dendritic cells or macrophages may improve antigen presentation and antitumor immunity, whereas persistent activation can support angiogenesis, suppressive myeloid recruitment, immune escape or tissue remodeling. Therefore, inflammasome findings should be interpreted through cellular compartment and chronicity, not only through sensor identity.

NLRP3 remains the best characterized inflammasome and consists of the NLRP3 sensor, ASC and pro-caspase 1 [48]. NLRP3 contains an N-terminal pyrin domain for ASC recruitment, a central NACHT domain that mediates ATP-dependent oligomerization, a C-terminal leucine-rich repeat domain involved in autoregulation and stimulus responsiveness [49–51]. Upon activation, ASC bridges NLRP3 and pro-caspase 1 through homotypic PYD-PYD and CARD-CARD interactions, allowing caspase 1 activation and subsequent GSDMD and cytokine processing. In tumors, NLRP3 is best viewed as a microenvironmental amplifier that can receive signals from dying cancer cells, damaged mitochondria, therapy-induced DAMPs and microbiota-related cues. This interpretation is more cancer-specific than simply stating that NLRP3 activates caspase 1, because the biological output depends on whether NLRP3 signaling occurs in malignant cells, macrophages, dendritic cells, neutrophils or stromal compartments.

Autophagy is not simply another parallel death modality in the cell-death network. It also functions as a quality-control system that limits endogenous inflammasome stimuli by removing damaged mitochondria, oxidized mitochondrial DNA, intracellular microbes, damaged ER membranes and aggregated proteins. When mitophagy or broader autophagic flux is impaired, damaged mitochondria accumulate, mitochondrial ROS and cytosolic mitochondrial DNA increase, and NLRP3 inflammasome activation is amplified [52, 53]. Autophagy can also restrain IL-1β output by targeting pro-IL-1β for lysosomal degradation [54].

In tumors, this regulatory layer has two implications. First, therapy-induced organelle stress may create DAMPs that prime myeloid inflammasomes even when the primary dying cell is apoptotic. Second, autophagy modulation can alter whether apoptotic chemotherapy ultimately remains non-inflammatory or progresses toward GSDME-dependent secondary pyroptosis and cytokine release. Thus, autophagy should be interpreted as a rheostat that controls the availability of pyroptosis triggers rather than as a detached topic from the apoptosis-pyroptosis switch.

Crosstalk between apoptosis and pyroptosis

The key cancer-relevant question is not whether apoptosis and pyroptosis are different, but how cells choose between them under therapeutic and immune pressure. This choice is governed by at least five layers: the initiating stimulus, the active caspase repertoire, the expression and cleavage site of gasdermins, the efficiency of efferocytic clearance, and the composition of the tumor microenvironment. A tumor cell with active caspase 3 but little GSDME will usually display apoptosis, whereas a GSDME-high cell exposed to the same upstream apoptotic stimulus can become lytic and inflammatory. Similarly, a death-receptor signal can remain apoptotic, switch to GSDMC-mediated pyroptosis, or be diverted toward necroptosis depending on caspase 8 activity, cFLIP levels and RIPK3-MLKL availability. Thus, apoptosis-pyroptosis crosstalk is best understood as a cancer-specific decision process that links tumor-intrinsic stress programs to immune-cell interpretation of dying cells.

Mechanistic integration: a cancer-specific switching framework

A useful way to integrate the evidence is to separate the switch into input, processing, execution, compartment and output layers. The input layer includes chemotherapy-induced DNA damage, death-receptor ligands, hypoxia, mitochondrial or ER stress, microbial products and DAMPs released from injured tissue. The processing layer is dominated by caspase 3, caspase 8, inflammatory caspases, cFLIP, IAPs and RIPK3/MLKL. The execution layer is defined by whether GSDME, GSDMC, GSDMD or GSDMB is expressed at sufficient protein abundance and cleaved at a pore-forming site. The compartment layer asks where the event occurs: malignant epithelial cells, infiltrating macrophages, dendritic cells, neutrophils, endothelial cells, stromal cells or cytotoxic lymphocyte targets. The output layer then determines clinical meaning: silent clearance, acute immunogenic inflammation, chronic tumor-promoting inflammation, vascular or tissue injury, or improved sensitivity to immunotherapy.

This layered framework prevents two common interpretive errors. First, it avoids treating infection-model or macrophage inflammasome mechanisms as if they automatically describe tumor-cell-intrinsic pyroptosis. Second, it avoids assuming that pyroptosis induction is always therapeutic. The same gasdermin pore formation that exposes tumor antigens and attracts immune effector cells can also release IL-1β, IL-18 and HMGB1 in a persistent pattern that supports myeloid recruitment, angiogenesis or inflammation-associated tumorigenesis. Therefore, the translational question is not simply whether pyroptosis can be induced, but whether the intended tumor compartment has the right caspase-gasdermin wiring and whether the surrounding microenvironment will convert lytic death into antitumor immunity rather than harmful inflammation.

Caspases as switches in apoptosis-pyroptosis crosstalk

Caspases constitute the central enzymatic hub linking apoptosis and pyroptosis. Initiator caspases such as caspase 8 and caspase 9 determine pathway entry, whereas executioner caspase 3 can either complete classical apoptosis or, in GSDME-expressing cells, convert apoptosis into pyroptosis (Fig. 3). This dual function is central to cancer therapy because many anticancer treatments were designed to activate apoptosis, yet their inflammatory output depends on substrate availability. Caspase nomenclature alone is therefore insufficient for cancer interpretation: caspase 3 is apoptotic when its relevant substrates are PARP, cytoskeletal proteins and DNA-repair proteins, but it becomes a pyroptosis trigger when GSDME is present; caspase 8 is apoptotic at the DISC, inflammatory when it cleaves gasdermins or regulates inflammasomes, and anti-necroptotic when it restrains RIPK3/MLKL. The biological consequence is determined by the tumor-cell substrate landscape and by whether nearby immune cells amplify or resolve the inflammatory signals.

Fig. 3.

Caspases-controlled apoptosis-pyroptosis switching. Caspase 3 connects apoptosis to GSDME-dependent secondary pyroptosis, while caspase 8 can support apoptosis, inflammasome signaling, gasdermin cleavage or necroptosis restraint depending on context

Several observations from infection and myeloid-cell models illustrate caspase plasticity. During Yersinia infection, caspase-8-dependent signaling contributes to caspase-1 processing and innate defense [55]. In caspase-1-deficient macrophages, AIM2/ASC-associated caspase 8 can support apoptosis rather than canonical pyroptosis [36, 40]. NAIP-NLRC4 inflammasome studies further show that caspase 8 can compensate for absent caspase 1 to sustain IL-18 release in some epithelial or myeloid contexts [56]. These findings are mechanistically informative, but they should be treated as context-defining examples, not as direct proof that the same wiring operates universally in tumor cells.

To make this evidence boundary explicit, four modules should be interpreted according to the model in which they were demonstrated. For caspase-8, infection and immune-cell studies show that caspase-8 can compensate for or modulate inflammasome signaling, whereas cancer-cell evidence is strongest when a tumor-specific upstream state and a gasdermin substrate are both shown, such as hypoxia/PD-L1-driven GSDMC expression followed by TNFα-caspase-8 cleavage or DR6/α-ketoglutarate-linked caspase-8/GSDMC activation. For inflammasome activation, macrophage or dendritic-cell NLRP3/AIM2/ASC-caspase-1 activity mainly indicates tumor-microenvironment cytokine output unless gasdermin cleavage and pore formation are demonstrated in malignant cells. For GSDME/GSDMC switching, the key tumor-cell evidence is substrate-specific: caspase-3 cleaves GSDME downstream of apoptotic therapy, whereas caspase-8 cleaves GSDMC in defined hypoxic, PD-L1-positive or DR6-linked cancer contexts. For granzyme-induced gasdermin cleavage, the cytotoxic lymphocyte supplies the granzyme, but the pyroptotic target-cell output depends on whether the malignant cell expresses the corresponding cleavable gasdermin substrate, such as GSDMB; this should not be interpreted as pyroptosis occurring primarily in lymphoid cells.

The most direct apoptosis-to-pyroptosis switch in cancer is the caspase-3/GSDME axis, which should be understood as an execution-substrate switch downstream of otherwise apoptotic caspase 3. DNA-damaging chemotherapy, mitochondrial apoptosis, death-receptor signaling and other therapeutic stresses can all converge on caspase 3 activation. In GSDME-low cells, active caspase 3 mainly cleaves classical apoptotic substrates such as DNA-repair, cytoskeletal and survival proteins, producing cell shrinkage, blebbing and non-lytic clearance. In GSDME-expressing cells, the same active caspase 3 cleaves human GSDME/DFNA5 at Asp270 in the interdomain linker, releasing the N-terminal pore-forming fragment from C-terminal autoinhibition [57, 58]. GSDME-N then oligomerizes and inserts into cellular membranes, generating pores that drive iron flux, water influx, cellular swelling, membrane rupture and release of DAMPs, cytokines and tumor antigens. Thus, caspase 3 converts an apoptotic protease signal into a gasdermin-executed inflammatory output when sufficient GSDME substrate is available.

This transition is controlled by several mechanistic thresholds. First, the magnitude and duration of caspase 3 activation determine how much GSDME-N is generated. Second, GSDME abundance and epigenetic state set the execution threshold: GSDME-high cells can undergo rapid pyroptotic lysis after caspase 3 activation, GSDME-low cells may first display apoptotic morphology and then progress to secondary pyroptosis, and GSDME-silenced cells remain predominantly apoptotic [57–60]. Third, subcellular amplification determines whether the switch becomes self-reinforcing. GSDME pores can permeabilize mitochondria, promote cytochrome c release and further augment caspase 3 activation, creating a feed-forward apoptosis-pyroptosis loop [61]. In tumors, this explains why chemotherapy-induced caspase 3 activation may be immunologically quiet in GSDME-silenced cancer cells but lytic, DAMP-releasing and immune-stimulatory in GSDME-competent cells; in mouse tumor models, GSDME expression suppressed tumor growth through cytotoxic lymphocyte-dependent antitumor immunity [62].

Caspase 8 is another major switching node, but its cancer relevance depends strongly on the evidence context. In classical extrinsic apoptosis, death receptors recruit FADD and activate caspase 8, which then activates caspase 3. In infection or innate-immune models, caspase 8 can also cleave GSDMD or GSDME and promote inflammatory death, particularly when TAK1 signaling is inhibited or pathogen effectors block NF-κB/MAPK survival pathways [55, 63, 64]. These studies establish the plasticity of caspase-8 signaling but do not by themselves prove tumor-cell pyroptosis. In cancer cells, the PD-L1/p-STAT3/GSDMC axis provides a direct tumor-specific example: hypoxia promotes nuclear PD-L1-associated GSDMC expression, and TNFα-activated caspase 8 then cleaves GSDMC, switching TNFα-induced apoptosis to pyroptosis [65]. A related cancer context is the α-ketoglutarate/DR6/caspase-8/GSDMC axis [66]. Thus, caspase-8-related conclusions should specify whether the evidence comes from infection/immune-cell inflammasome models or from tumor-cell systems in which GSDMC or another gasdermin substrate is directly demonstrated.

Caspase 8 also shapes inflammasome output, but this statement needs compartment-specific interpretation. In dendritic cells, macrophages or infection models, caspase 8 can modulate IL-1β secretion, participate in noncanonical inflammasome complexes and regulate caspase 1 or caspase 11 activation depending on the stimulus [36, 40, 67, 68]. In tumors, these findings most directly support a role for immune-cell or myeloid inflammasome activity within the tumor microenvironment. They should be distinguished from direct tumor-cell pyroptosis, which requires evidence that malignant cells contain the relevant inflammasome or caspase machinery, cleave GSDMD or another gasdermin, and form pore-generating GSDM-N fragments. This distinction is important for therapy response because chemotherapy or dying-cell DAMPs may activate inflammasomes in infiltrating myeloid cells even when the primary tumor-cell death is apoptotic.

The canonical and noncanonical pyroptotic routes have distinct roles in cancer progression and therapy response. Canonical inflammasome signaling is usually initiated when sensors such as NLRP3, AIM2, NLRC4 or NLRP1 integrate DAMPs, mitochondrial injury, ATP, ROS, microbial cues or therapy-induced stress into ASC-caspase-1 activation. In cancer progression, this route often acts as a tumor-microenvironment amplifier rather than a purely tumor-cell-intrinsic death program. Short-lived canonical activation in dendritic cells or macrophages may support antigen presentation, IL-1β/IL-18 maturation and recruitment of cytotoxic lymphocytes, whereas persistent canonical inflammasome activity can maintain myeloid inflammation, angiogenesis, stromal remodeling, epithelial barrier damage and immune suppression. For therapy response, canonical inflammasome activation can therefore be beneficial when it converts local tumor damage into acute antitumor immunity, but it can be harmful when chemotherapy, radiotherapy or dying-cell DAMPs create chronic IL-1β/IL-18 loops, normal-tissue injury or tumor-supportive inflammation. For example, the anthracyclines doxorubicin and daunorubicin can induce NLRP3-dependent processing and release of IL-1β, illustrating that cytotoxic therapy may engage canonical inflammasome output in addition to direct apoptotic tumor-cell killing [69].

By contrast, noncanonical caspase-4/5/11-GSDMD signaling is triggered most directly by cytosolic LPS and is therefore especially relevant to tumors or treatment settings associated with microbial translocation, barrier disruption, intracellular bacterial products or local endotoxin exposure. Its role in cancer progression is less about sterile DAMP sensing and more about linking microbiota-related signals to GSDMD pore formation, potassium efflux, secondary NLRP3 activation and inflammatory remodeling of the tumor microenvironment. For therapy response, the evidence is more context-dependent. In immune-cell models, proapoptotic chemotherapeutic drugs can induce noncanonical processing and release of IL-1β through caspase-8 in dendritic cells, suggesting that chemotherapy may produce inflammatory cytokine output through noncanonical inflammasome wiring in the tumor microenvironment [70]. In tumor cells, LPS-enhanced GSDMD-mediated pyroptosis increased oxaliplatin sensitivity in HT29 colorectal cancer cells, supporting a more direct noncanonical sensitization example [71]. These examples should still be interpreted carefully: caspase-8-dependent noncanonical IL-1β processing in dendritic cells is not identical to the classical cytosolic LPS-caspase-4/5/11-GSDMD axis, and both routes can intensify cytokine release and tissue injury. Thus, canonical signaling is best interpreted as a DAMP/inflammasome-caspase-1 cytokine axis that can either activate or chronically distort the tumor microenvironment, whereas noncanonical signaling is a cytosolic LPS- or caspase-8-linked inflammatory axis whose therapeutic relevance depends strongly on cell type, barrier status, microbial exposure and local inflammatory control.

Finally, caspase 8 helps position apoptosis-pyroptosis switching within the broader regulated-death network. Its catalytic activity suppresses RIPK3-MLKL-dependent necroptosis, and reduced caspase 8 activity can divert death-receptor or pattern-recognition receptor signaling toward necroptosis or PANoptosis-like mixed death states [7, 72–75]. This explains why apoptosis, pyroptosis and necroptosis often behave as interconnected nodes rather than discrete boxes. Ferroptosis can also intersect indirectly by increasing ROS, membrane damage and DAMP release, thereby changing inflammasome tone in the tumor microenvironment. These intersections should be interpreted as context-dependent regulatory networks rather than as interchangeable death labels.

Gasdermins as execution switches

Gasdermins provide the execution layer of pyroptosis and are therefore decisive molecular switches in the apoptosis-pyroptosis interface (Fig. 4). The human family includes GSDMA, GSDMB, GSDMC, GSDMD, GSDME/DFNA5 and pejvakin (PJVK/DFNB59) [76]. Most gasdermins share an autoinhibited architecture in which a C-terminal domain restrains an N-terminal pore-forming domain until proteolytic cleavage releases the active fragment. Their cancer roles differ by cell type and clinical context: GSDMD is the canonical inflammatory-caspase substrate and is often prominent in myeloid inflammation; GSDME links apoptotic caspase 3 to tumor-cell pyroptosis; GSDMC can mediate hypoxia/PD-L1/TNFα-driven switching; and GSDMB can be cleaved by granzyme A in target cells recognized by cytotoxic lymphocytes [57, 62, 65, 76, 77]. For cancer-specific interpretation, gasdermin expression should be treated as a functional biomarker rather than a pathway label, because the same caspase activation can be non-lytic, lytic, immunogenic or toxic depending on which gasdermin is expressed in which compartment.

Fig. 4.

Gasdermin-dependent execution switches. GSDMD, GSDME, GSDMC and GSDMB convert caspase or granzyme activity into apoptosis, direct pyroptosis or secondary pyroptosis depending on expression and cleavage context

Cleavage site and expression level are both critical. In canonical pyroptosis, inflammatory caspases cleave human GSDMD at Asp275 to generate a lytic N-terminal fragment [78, 79]. By contrast, apoptotic caspase 3 can cleave GSDMD at a different site that generates a non-lytic fragment, thereby favoring apoptosis rather than pyroptosis [8]. For GSDME, the amount of available substrate determines the morphology following caspase 3 activation: GSDME-null tumor cells undergo apoptosis; GSDME-high cells rapidly undergo pyroptosis; and cells with lower GSDME may show an apoptotic phase before progressing to secondary pyroptosis [57, 58]. GSDME expression is frequently controlled by epigenetic state and cell-state programs such as epithelial-mesenchymal transition, suggesting that DNMT inhibition, transcriptional context and tumor differentiation state may alter pyroptotic competence [60].

GSDME-reactivating strategies should therefore be interpreted as attempts to restore this execution-substrate switch rather than merely as ways to increase total cell death. The mechanistic premise is that many tumor cells still activate caspase 3 after chemotherapy, targeted stress or death-receptor input, but the absence or epigenetic silencing of GSDME makes this death relatively apoptotic and immunologically quiet. Restoring GSDME expression changes the substrate available to caspase 3: the same upstream apoptotic signal can now produce GSDME-N pores, membrane rupture, DAMP release, secondary pyroptosis and a more immunogenic tumor-cell death phenotype. This conversion is therapeutically meaningful only when it is tumor-selective enough to increase cancer-cell killing and immune activation without causing uncontrolled normal-tissue inflammation.

GSDMC and GSDMB extend this switch beyond the caspase-3/GSDME axis, but they also illustrate why the cellular source of the evidence must be stated. For GSDMC, direct cancer-cell evidence links hypoxia, nuclear PD-L1 and p-Y705-STAT3 to GSDMC transcription; when TNFα activates caspase 8, GSDMC is cleaved at 362LELD365 and its N-terminal fragment executes pyroptosis [65]. This is different from infection-model caspase-8 plasticity because the tumor-cell gasdermin substrate and cancer-associated upstream state are both defined. For GSDMB, granzyme A released by cytotoxic lymphocytes can cleave GSDMB at Lys244, allowing natural killer cells and cytotoxic T lymphocytes to induce inflammatory death in GSDMB-expressing target cells [77]. Here, the lymphocyte provides the protease, whereas the target cell provides the gasdermin substrate; therefore, granzyme-induced gasdermin cleavage should be interpreted as immune-cell-mediated killing of gasdermin-competent target cells rather than as evidence that lymphoid cells are the main pyroptotic compartment. These mechanisms are attractive for cancer immunotherapy, but if gasdermin expression is mainly stromal, myeloid or normal epithelial rather than malignant, the same strategy may intensify tissue injury or chronic inflammation without improving tumor-cell eradication.

cFLIP, death receptors and granzyme pathways

Cellular FLICE-like inhibitory protein (cFLIP) regulates the death-receptor complex and therefore acts as a rheostat for apoptosis, necroptosis and pyroptosis (Fig. 5) [80, 81]. High cFLIP(L) can restrain caspase 8 activation at the DISC and suppress downstream inflammatory death, whereas reduced cFLIP(L) permits caspase 8 activation and can promote noncanonical NLRP3 inflammasome activation and IL-1β release in LPS-primed macrophages [82–85]. Complete cFLIP(L) deficiency can free caspase 8 to cleave GSDMD or activate the NLRP3 inflammasome, but this mechanism has been best defined in immune-cell systems. In cancer, cFLIP is also a well-known apoptosis-resistance factor. Therefore, cFLIP-targeting strategies cannot be interpreted by apoptosis restoration alone: depending on RIPK3/MLKL status, gasdermin expression and myeloid priming, the same intervention may restore apoptotic sensitivity, promote inflammatory pyroptosis, or shift cells toward necroptosis-like death.

Fig. 5.

cFLIP, death-receptor and granzyme regulation. cFLIP tunes caspase 8 activity, while Fas-FasL and perforin/granzyme pathways connect immune-cell killing to apoptosis or gasdermin-mediated pyroptosis

The Fas-FasL axis illustrates a similar context dependence. In tumor cells, Fas signaling can assemble the DISC and trigger caspase-8/caspase-3-mediated apoptosis [85]. In TLR-primed macrophages and dendritic cells, however, Fas upregulation can drive caspase-8-dependent maturation and release of IL-1β and IL-18 while bypassing canonical inflammasomes and RIPK3/MLKL-mediated necroptosis [86]. Thus, the same ligand-receptor axis may contribute to tumor-cell killing, inflammatory cytokine release or immune-cell reprogramming, depending on the cellular compartment in which it is activated.

The perforin/granzyme pathway further connects apoptosis and pyroptosis to antitumor immunity. Granzyme B classically activates caspase-dependent apoptosis in target cells, whereas granzyme A can cleave GSDMB to trigger lytic pyroptosis [77, 87]. These findings broaden the effector repertoire of natural killer cells and cytotoxic T lymphocytes beyond non-inflammatory apoptosis. In therapeutic terms, they suggest that gasdermin expression in tumor cells may influence the quality of immune-mediated killing, not only the quantity of tumor-cell death. They also suggest a cancer-specific biomarker question: immune infiltration will not necessarily produce pyroptotic tumor killing unless the target malignant cells express a cleavable gasdermin substrate.

Discussion

Integrated synthesis of apoptosis-pyroptosis switching in cancer

This framework supports a more precise conclusion: apoptosis and pyroptosis are not parallel destinations but context-dependent outputs of shared signaling machinery. Caspase-3, caspase-8, inflammatory caspases, granzymes and gasdermins form the central switching module, while cFLIP, IAPs, RIPK1/RIPK3/MLKL, autophagy and inflammasome priming define the permissive or restrictive environment. It is therefore more accurate to state that some inflammasome-associated or caspase-associated lytic processes may occur when specific gasdermins are absent, whereas canonical pyroptosis requires gasdermin pore formation.

The cancer-specific meaning of this framework is that pathway names alone do not predict biological output. Chemotherapy or DNA damage often enters through mitochondrial apoptosis, but in GSDME-positive tumor cells the same caspase-3 signal can become a pore-forming GSDME-N event. Death-receptor or TNF-related signaling may remain extrinsic apoptosis, be restrained by cFLIP, connect to RIPK1/RIPK3/MLKL-dependent inflammatory death, or switch to caspase-8/GSDMC-mediated pyroptosis in hypoxic, PD-L1-high or DR6/alpha-ketoglutarate-linked cancer contexts. Similarly, DAMPs, PAMPs and defective autophagic clearance mainly shape inflammasome-GSDMD signaling in myeloid-rich tumor microenvironments, but the resulting cytokine and DAMP field can feed back onto tumor cells, stromal cells and immune recruitment.

The practical implication is that apoptosis-pyroptosis crosstalk should be interpreted through a sequence of cancer-specific questions: which stress initiated the response, which protease route is active, which gasdermin substrate is present at the protein level, which cellular compartment undergoes lytic death, and whether the output is acute tumor-cell elimination, immune activation, immune suppression or tissue-damaging inflammation. This logic also explains why a single therapeutic agent can have different consequences across cancers: the same chemotherapy, death-receptor agonist, oncolytic virus, epigenetic drug or immune effector pathway may produce silent apoptosis in one tumor, GSDME- or GSDMC-dependent tumor-cell pyroptosis in another, or predominantly myeloid inflammasome activation in a third.

This synthesis also explains why the therapeutic goal should be control of the switch rather than indiscriminate pyroptosis induction. When gasdermin cleavage is acute, tumor-localized and paired with antigen presentation, it may overcome apoptotic resistance and enhance antitumor immunity. When the same lytic program occurs in normal epithelial, endothelial, stromal or myeloid compartments, or when cytokine release is prolonged, it may drive excessive inflammation, collateral tissue injury, suppressive myeloid recruitment and chronic tumor-supportive inflammation. Thus, the central translational question is not whether pyroptosis can be induced, but whether the apoptosis-to-pyroptosis switch can be directed to the right cell type, at the right intensity and for the right duration.

Autophagy in apoptosis-pyroptosis crosstalk and DAMP/PAMP control

Autophagy needs to be interpreted within both apoptosis and pyroptosis because it changes how tumor cells and the tumor microenvironment respond to stress. In the apoptotic context, basal autophagy removes damaged mitochondria and ER components, limits ROS accumulation and can raise the threshold for mitochondrial outer membrane permeabilization and caspase-3 activation. When autophagy is defective or overwhelmed by chemotherapy, hypoxia, nutrient deprivation or ER stress, damaged organelles accumulate and apoptotic signaling may be amplified. In the pyroptotic context, autophagy restrains inflammasome priming by removing the endogenous and microbial danger signals that would otherwise activate inflammatory caspases and gasdermin-dependent pore formation. Thus, autophagy is not a detached cell-death topic; it is a regulatory checkpoint that helps determine whether therapy-induced stress is resolved as relatively silent apoptosis, converted into GSDME-dependent secondary pyroptosis, or amplified through inflammasome-driven tumor microenvironment inflammation.

This regulatory connection is particularly important for DAMP/PAMP control. Autophagy can remove intracellular bacteria, damaged mitochondria, oxidized mitochondrial DNA, damaged ER membranes and protein aggregates, all of which can otherwise function as inflammasome stimuli [52–54]. If these substrates are not cleared, they can promote NLRP3 or related inflammasome activation in macrophages, dendritic cells or other tumor microenvironment populations, increase IL-1β/IL-18 release, and intensify DAMP amplification after tumor-cell death. In cancer therapy, this means that autophagy status may influence whether apoptosis-pyroptosis crosstalk remains controlled and immunogenic or progresses toward excessive inflammatory toxicity and chronic tumor-supportive inflammation.

It is also important not to overstate the speed or energetic logic of pyroptosis. Pyroptosis can appear rapidly after gasdermin cleavage because pore formation immediately disrupts membrane integrity, but speed depends on stimulus, cell type, gasdermin abundance, repair capacity and phagocytic clearance [57]. There is currently insufficient evidence to claim that tumor cells preferentially choose pyroptosis to conserve energy for surviving cancer cells. The more defensible point is that therapy-activated caspases may produce different inflammatory consequences depending on available gasdermin substrates.

Therapeutic opportunities and resistance

A major translational opportunity is to use pyroptosis to overcome drug resistance, but the therapeutic examples should be interpreted as apoptosis-pyroptosis crosstalk-node interventions rather than as a general instruction to induce inflammation. At the caspase-3/GSDME node, chemotherapy-induced apoptosis becomes pyroptotic only when GSDME is expressed and cleavable [57]; therefore, GSDME-priming strategies such as the DNMT inhibitor decitabine in resistant MCF-7/Taxol breast cancer cells, decitabine or GSDME overexpression in retinoblastoma cells, and simvastatin combined with the DNMT inhibitor 5-Aza-2’-deoxycytidine in gastric cancer models aim to restore the substrate that allows caspase 3 to execute pyroptosis [88–90]. At the BAX/caspase-3/GSDME node, GW4064 combined with oxaliplatin enhanced colorectal cancer chemosensitivity by inducing pyroptosis [91]. At the caspase-8/GSDMC node, hypoxia/PD-L1/TNFα signaling and α-ketoglutarate/DR6 signaling redirect death-receptor or stress-linked caspase-8 activity toward GSDMC-dependent lytic death [65, 66]. At the GSDMD node, LPS enhanced oxaliplatin sensitivity in HT29 cells through GSDMD-mediated pyroptosis, linking inflammatory sensing to chemotherapy response [71]. At the immune-combination node, oncolytic vesicular stomatitis virus-induced GSDME pyroptosis and GSDMB-N mRNA lipid nanoparticles improved responsiveness to immune-checkpoint blockade in preclinical models [92, 93]. These examples make the translational principle more specific: the drug, inhibitor or experimental therapy should match an upstream apoptotic or inflammatory trigger to the relevant caspase, gasdermin substrate and tumor-immune context.

GSDME reactivation can be developed as a two-step therapeutic strategy. The first step is priming: identifying tumors in which GSDME is methylated, transcriptionally suppressed or expressed below the threshold needed for pore-forming pyroptosis, and then restoring GSDME through epigenetic or transcriptional intervention. The second step is triggering: combining this priming with a chemotherapy, targeted agent, death-receptor stimulus or immune effector pathway that activates caspase 3 or granzyme B in tumor cells. In this design, the goal is not simply to increase apoptosis, but to change the final execution route from silent apoptotic clearance to inflammatory GSDME-mediated pyroptosis.

The drug-sensitivity rationale has several layers. First, GSDME-N pores provide a terminal lytic output when apoptotic execution is incomplete or delayed, thereby lowering the amount of cytotoxic stress needed to kill tumor cells. Second, GSDME pores can amplify mitochondrial damage and caspase 3 activation, reinforcing the apoptosis-pyroptosis loop rather than leaving caspase activation as a reversible or sublethal signal [61]. Third, experimental studies show that this is not only theoretical: decitabine-mediated GSDME demethylation increased Taxol sensitivity in resistant MCF-7/Taxol breast cancer cells, decitabine or GSDME overexpression increased carboplatin sensitivity in retinoblastoma cells, and simvastatin combined with the DNA methyltransferase inhibitor 5-Aza-2’-deoxycytidine promoted GSDME-mediated pyroptosis in gastric cancer models [88–90].

The anti-tumor immunity rationale is equally important. GSDME-mediated pyroptosis changes the immune visibility of dying tumor cells by releasing DAMPs, cytokines and tumor antigens, and by creating a local inflammatory context that can improve dendritic-cell priming, macrophage phagocytosis and NK-cell or CD8 + T-cell recruitment. In mouse tumor models, GSDME suppressed tumor growth through cytotoxic lymphocyte-dependent antitumor immunity and enhanced tumor-cell phagocytosis by tumor-associated macrophages [62]. This provides the mechanistic basis for pairing GSDME-reactivating or GSDME-executing strategies with immune checkpoint blockade, oncolytic virus therapy or other immunostimulatory treatments, provided that the inflammatory output remains acute and tumor localized.

For translation, this emerging approach should therefore be evaluated with a biomarker-guided framework. Candidate tumors should be assessed for GSDME promoter methylation, GSDME protein abundance, caspase-3 competence, apoptotic drug responsiveness and the cellular compartment in which GSDME is expressed. Treatment scheduling may also matter: an epigenetic priming phase may be needed before the cytotoxic trigger, whereas checkpoint blockade or other immune partners may be most useful after pyroptotic antigen and DAMP release has converted an immune-cold lesion into a more inflamed state. At the same time, this strategy should not be generalized to every cancer, because systemic demethylation, GSDME expression in normal epithelia and chronic IL-1β/IL-18-driven inflammation could narrow the therapeutic window.

In combination therapies aimed at boosting anti-tumor immune responses, pyroptosis can function as an immune-sensitizing death-reprogramming module. One component can provide the tumor-cell stress or gasdermin-executing trigger, whereas the partner therapy amplifies immune recognition or removes immune brakes. For example, oncolytic vesicular stomatitis virus can induce GSDME-dependent tumor-cell pyroptosis and sensitize immunologically cold tumors to checkpoint blockade [92], while mRNA lipid nanoparticles encoding the N-terminal domain of GSDMB can directly trigger tumor pyroptosis and improve anti-PD-1 responses in 4T1 and B16F10 models [93]. Thus, pyroptosis may help overcome resistance not only by killing tumor cells that evade apoptosis, but also by converting resistant or immune-cold lesions into an inflammatory state that can be amplified by checkpoint blockade or other immunostimulatory combinations.

Pyroptosis may also convert immunologically cold tumors into hot tumors by releasing DAMPs, IL-1β, IL-18 and tumor antigens that recruit and activate dendritic cells, natural killer cells and cytotoxic T lymphocytes. GSDME has been shown to suppress tumor growth through antitumor immunity [62]. From a translational perspective, this means that the most useful pyroptosis-based regimens are likely to be biomarker-guided combinations: the tumor should retain a cleavable gasdermin execution substrate, the treatment should activate the matching caspase or granzyme route, and the immune partner should be chosen to expand the resulting acute antitumor response rather than merely adding another cytotoxic pressure.

Balanced therapeutic development must also consider harm. Pyroptosis induction is not automatically beneficial. If activation is excessive, poorly localized or chronic, the release of IL-1β, IL-18, HMGB1, ATP and other DAMPs may damage normal tissue, recruit suppressive myeloid cells or create tumor-supportive inflammation. In colitis-associated colorectal cancer, GSDME-mediated pyroptosis of epithelial cells promoted tumorigenesis through HMGB1/ERK1/2 signaling [94]. Therefore, productive pyroptosis therapy should answer four linked questions before clinical translation: which tumor cells express the relevant gasdermin protein; which caspase can cleave it under the planned treatment; which immune or stromal cells will sense the released cytokines and DAMPs; and whether the expected inflammatory output is acute antitumor immunity or chronic tissue-damaging inflammation.

Biomarkers, single-cell resolution and limitations

Because pyroptosis is highly cell-type specific, bulk expression of a gasdermin or inflammasome component is insufficient for patient stratification. A high pyroptosis-related gene score in a whole tumor may reflect tumor-intrinsic GSDME/GSDMC competence, inflammasome-primed macrophages, dendritic cells or neutrophils, or stromal/endothelial inflammatory responses. Single-cell RNA sequencing therefore changes the biomarker question from whether a tumor expresses pyroptosis genes to which cellular compartment can execute pyroptosis and which compartment mainly responds to pyroptotic products. In practical terms, malignant cells should be evaluated for gasdermin execution substrates, caspase-3/8 activity and epigenetic silencing of GSDME, whereas myeloid cells should be evaluated for NLRP3/ASC/caspase-1, IL1B/IL18 and GSDMD modules. Lymphoid populations may show lower direct pyroptotic activity but can be decisive responders through granzyme release, IFN signaling, checkpoint expression and cytotoxic recruitment.

At cellular resolution, pyroptosis can shape the tumor microenvironment through at least three linked routes. First, tumor-cell pyroptosis can release DAMPs, tumor antigens and cytokines, supporting dendritic-cell maturation, macrophage phagocytosis and recruitment or activation of CD8 + T cells and NK cells. Second, inflammasome-high myeloid cells can amplify IL-1β/IL-18 and chemokine circuits, which may enhance acute antitumor immunity but may also recruit suppressive myeloid cells, maintain tumor-supportive chronic inflammation, stimulate angiogenesis or damage adjacent tissue. Third, stromal and endothelial responses can translate pyroptotic signals into extracellular-matrix remodeling, vascular activation and ligand-receptor communication with malignant cells. Thus, the same bulk pyroptosis signature may represent therapeutically favorable tumor-cell lysis or an inflammatory myeloid niche that supports progression, depending on the cellular source and temporal context.

Recent scRNA-seq studies illustrate this point. In colorectal cancer, bulk and single-cell analyses linked pyroptosis-related risk scores with immune infiltration and predicted sensitivity to 5-fluorouracil and anti-PD-1 therapy, and single-cell analysis localized pyroptosis-related genes within the tumor ecosystem [95]. In uterine corpus endometrial carcinoma, scRNA-seq revealed cell-type-specific pyroptosis regulators across epithelial cells, macrophages, endothelial cells, fibroblasts and lymphocytes, and suggested that dysregulated pyroptosis-related genes alter cell-cell communication and TNF-pathway ligand-receptor interactions in the tumor microenvironment [96]. Most directly related to the cell-type specificity raised in this comment, a lung adenocarcinoma study integrating scRNA-seq and bulk analyses identified 17 cell types and reported higher pyroptosis-related activity in myeloid dendritic cells and neutrophils than in cancer cells, with enrichment of inflammasome assembly, NOD-like receptor signaling and immune pathways [97].

A practical conclusion from these studies is that scRNA-seq should not be used only to generate another prognostic signature. It should guide a compartment-resolved therapeutic decision: whether to activate GSDME/GSDMC-dependent pyroptosis in malignant cells, restrain tumor-supportive inflammasome activity in myeloid cells, or combine pyroptosis induction with immune-checkpoint blockade when lymphoid effector programs are present. However, scRNA-seq measures transcripts rather than pore formation or cell lysis. Therefore, single-cell findings should be paired with spatial transcriptomics or multiplex immunostaining for GSDMD-N/GSDME-N, cleaved caspase-1/3, IL-1β/IL-18, HMGB1 or LDH release, so that the inferred pyroptotic compartment is confirmed at protein, spatial and functional levels.

Several limitations and translational challenges remain, and the most clinically important concern is whether pyroptosis can be restricted to malignant cells without producing excessive inflammation or collateral injury to surrounding normal tissues. First, much of the mechanistic evidence still derives from macrophages, infection systems or forced gasdermin overexpression, which does not establish whether a clinically deliverable regimen can selectively activate gasdermin pores in tumor cells while sparing normal epithelial, endothelial, hepatic, intestinal, stromal and immune compartments. Second, pyroptosis has a narrow spatial and temporal therapeutic window: if gasdermin pores form outside the tumor-cell compartment, or if lytic death is too extensive or prolonged, IL-1β, IL-18, HMGB1, ATP, mitochondrial DNA and LDH release may damage epithelial barriers, vascular beds and adjacent parenchyma before a useful antitumor immune response is established. Third, this inflammatory injury may itself reshape the tumor microenvironment. Repeated or poorly resolved pyroptosis can amplify stromal inflammation, recruit suppressive myeloid cells, promote angiogenesis and tissue remodeling, and maintain chronic tumor-supportive inflammatory states rather than acute tumor rejection; GSDME-mediated epithelial pyroptosis has been linked to colitis-associated colorectal tumorigenesis through HMGB1/ERK1/2 signaling [94]. Finally, current biomarkers are insufficient because gasdermin mRNA or pyroptosis scores do not prove GSDM-N pore formation, the compartment of lytic death, the duration of cytokine release, the degree of bystander inflammation or organ safety. Translation therefore requires integrated evidence of tumor-selective delivery, gasdermin protein localization, active caspase or granzyme route, GSDM-N fragments, acute versus chronic inflammatory output, LDH/cytokine release, barrier integrity and normal-tissue injury.

Conclusion

In conclusion, apoptosis-pyroptosis crosstalk connects apoptotic caspases, inflammatory caspases, granzymes and gasdermins into a context-dependent death-switching network in cancer. The final outcome is determined by the initiating stress, active protease route, gasdermin availability and cellular compartment, and may range from silent apoptosis to tumor-cell pyroptosis, antitumor immunity or harmful inflammation (Fig. 6). Defining these switching rules should help guide biomarker-informed cancer therapy while avoiding indiscriminate pyroptosis induction.

Fig. 6.

Integrated model of apoptosis-pyroptosis crosstalk in cancer. Chemotherapy, death-receptor signaling, hypoxia, microbial/LPS-related cues, autophagy status and CTL/NK-cell attack converge on caspase-, inflammasome- and granzyme-dependent processing nodes. Gasdermin availability and cleavage determine whether these inputs lead to GSDME-, GSDMC-, GSDMD- or GSDMB-mediated pyroptosis, silent apoptosis/efferocytosis, acute antitumor immunity or chronic inflammatory toxicity. Biomarker-guided therapy should consider gasdermin expression, active caspase route, cellular compartment, immune context, delivery, dose, timing and normal-tissue safety

Abbreviations

AIF

Apoptosis-inducing factor

AIM2

Absent in melanoma 2

ALR

AIM2-like receptor

Apaf-1

Apoptotic protease-activating factor 1

ASC

Apoptosis-associated speck-like protein containing a CARD

Bcl-2

B-cell lymphoma-2

CARD

Caspase activation and recruitment domain

cFLIP

Cellular FLICE-like inhibitory protein

CTLs

Cytotoxic T lymphocytes

DAMPs

Damage-associated molecular patterns

DCs

Dendritic cells

DISC

Death-inducing signaling complex

DPP8/9

Dipeptidyl peptidase 8/9

DR6

Death receptor 6

ER

Endoplasmic reticulum

FADD

Fas-associated death domain

FasL

Fas ligand

GSDM

Gasdermin

GSDMB

Gasdermin B

GSDMC

Gasdermin C

GSDMD

Gasdermin D

GSDME

Gasdermin E

HD

Helical domain

IAPs

Inhibitor of apoptosis proteins

IL-18

Interleukin-18

LPS

Lipopolysaccharide

LRR

Leucine-rich repeat

MLKL

Mixed lineage kinase domain-like protein

MOMP

Mitochondrial outer membrane permeabilization

NAIP

Neuronal apoptosis inhibitory protein

NBD

Nucleotide-binding domain

NK

Natural killer

NLR

NOD-like receptor

NLRC4

NLR family CARD domain-containing protein 4

NLRP3

NOD-like receptor protein 3

NPCD

Non-programmed cell death

PAMPs

Pathogen-associated molecular patterns

PCD

Programmed cell death

PD-L1

Programmed death-ligand 1

PRR

Pattern recognition receptor

PYD

Pyrin domain

RIG-I

Retinoic acid-inducible gene I

RIPK1/3

Receptor-interacting serine/threonine kinase 1/3

RLR

RIG-I-like receptor

ROS

Reactive oxygen species

Smac/DIABLO

Second mitochondria-derived activator of caspases

STAT3

Signal transducer and activator of transcription 3

TLR

Toll-like receptor

TME

Tumor microenvironment

TNF

Tumor necrosis factor

TNFR1

Type 1 tumor necrosis factor receptor

TRADD

TNF receptor-associated death domain

TRAF2

TNF receptor-associated factor 2

WHD

Winged helix domain

Authors’ contributions

Liu YF drafted, edited, and approved the manuscript, Xie Y provided useful suggestions about editing the manuscript.

Funding

This study was supported by Health Research Project of Hunan Provincial Health Commission (No. 20254783), ‌The Science and Technology Program of Hunan Province (No. 2026JJ70129, 2026JJ82240), Education Department of Hunan Province of China (No. 24C0205), Research Project of Changsha Central Hospital (No. YNKY202504).

Data availability

Not applicable.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.