Is metabolic dysfunction-associated steatotic liver disease truly irrelevant? Unmasking the heterogeneity and confounders in pancreatic cancer outcomes

Abstract

Metabolic dysfunction-associated steatotic liver disease (MASLD) has no significant impact on liver-metastatic patterns or survival outcomes in patients with pancreatic cancer. While the negative findings are clinically appealing, several issues may limit a definitive biological interpretation. First, the use of steatosis-weighted, noninvasive surrogates may be prone to misclassification in pancreatic cancer, where cachexia, sarcopenia, and systemic inflammation can distort these components, thereby attenuating true associations toward the null. Second, accumulating evidence has suggested that MASLD is highly heterogeneous, in which adverse oncologic outcomes are more consistently linked to metabolic inflammation and fibrosis rather than steatosis alone and pooling these phenotypes may dilute signals from high-risk subgroups. Third, in pancreatic cancer with short survival, death is a major competing event, conventional time-to-event analyses without competing-risk frameworks may thereby underestimate the metastasis-related effect. Additionally, mechanistic research supports the concept of a pre-metastatic hepatic niche that is driven by inflammation and metabolic reprogramming in pancreatic cancer. Therefore, we conclude that the statistical absence of correlation should not be conflated with the absence of biological causation. Instead, future studies incorporating fibrosis-focused phenotyping, treatment-aware modeling, and competing-risk analyses are warranted to clarify which MASLD subtypes may meaningfully influence metastasis and survival.

Key Words: Metabolic dysfunction-associated liver disease; Pancreatic cancer; Metastasis; Statistical biases; Phenotypic heterogeneity; Biological plausibility

Core Tip: This article challenges the conclusion that metabolic dysfunction-associated steatotic liver disease is irrelevant to pancreatic cancer outcomes. We argue that the reported lack of association stems from the use of noninvasive surrogates that fail to capture the complexity of metabolic inflammation in cachectic patients. By ignoring the heterogeneity of metabolic dysfunction-associated steatotic liver disease and the statistical impact of competing risks, current analyses likely underestimate the pro-metastatic potential of the fibrotic liver niche. Furthermore, by contrasting these limitations with mechanistic evidence of a pre-metastatic niche, we emphasize that statistical absence of correlation does not equal biological absence of causation, advocating for fibrosis-focused future research.

INTRODUCTION

Chon et al[1] addresses a pivotal and increasingly relevant clinical question regarding the intersection of metabolic health and oncologic outcomes. The authors evaluated the impact of metabolic dysfunction-associated steatotic liver disease (MASLD) on the patterns of liver metastasis and overall survival in a large cohort of patients with pancreatic ductal adenocarcinoma. Given the rising global prevalence of metabolic dysfunction and the notorious lethality of pancreatic cancer, the determination of whether the hepatic microenvironment serves as a fertile soil for metastatic dissemination is of paramount importance. The authors concluded that MASLD, as primarily defined by the hepatic steatosis index (HSI), exerts no significant impact on the patterns of liver metastasis or survival outcomes in this population.

While the conclusion that MASLD does not complicate the clinical prognosis is intuitively appealing, we contend that the reported absence of a statistical association is likely a reflection of the inherent limitations of steatosis-weighted surrogates when applied to a cachectic population, rather than evidence of true biological independence. The extrapolation of epidemiological tools designed for the general population to the complex physiology of pancreatic cancer presents unique challenges that may obscure underlying risks. Upon close examination of the methodology and the biological plausibility of the findings, several critical concerns emerge. These concerns range from the paradox of measuring metabolic health in cachectic patients to the phenotypic heterogeneity of liver disease, the statistical artifacts introduced by competing risks, and the deep mechanistic contradictions. Hence, this article aims to elucidate these complexities to ensure that the potential role of the hepatic niche in the progression of pancreatic cancer is not prematurely dismissed.

THE PARADOX OF NONINVASIVE SURROGATES IN A CATABOLIC DISEASE CONTEXT

The primary methodological concern lies in the reliance on the HSI as the defining metric for MASLD in the large-scale primary analysis. The HSI, which incorporates the body mass index, the ratio of alanine aminotransferase to aspartate aminotransferase, and the presence of diabetes mellitus, has indeed proven to be a reliable surrogate for diagnosing hepatic steatosis in the general, non-cancer population[2,3]. However, the application of this index in patients with pancreatic cancer is compromised by the unique pathophysiology of the disease[4]. Pancreatic cancer is distinctively characterized by profound metabolic dysregulation, including cachexia, sarcopenia, and systemic inflammation, all of which directly distort the parameters used in the diagnostic formula[5]. Although the HSI has not yet been directly validated in cohorts specifically comprising patients with pancreatic cancer, evidence from broader oncologic populations indicates that cancer cachexia is already prevalent at the time of diagnosis in this malignancy and is typically accompanied by an average reduction of approximately 5%-20% relative to pre-morbid body weight[6,7]. Such a degree of weight loss is entirely sufficient to drive the HSI below its diagnostic cut-off, thereby potentially reclassifying individuals with pre-existing hepatic steatosis as metabolically normal.

The most striking confounder is the phenomenon of cancer-associated weight loss[8]. A significant proportion of patients with pancreatic cancer present with unintentional weight loss and skeletal muscle wasting at the time of diagnosis, which artificially depresses the body mass index[9,10]. Consequently, a patient with pre-existing MASLD who has lost significant weight due to cancer cachexia may present with a normalized body mass index, resulting in a score that falsely classifies the patient as free of MASLD. This introduces a systematic bias known as differential misclassification, where the patients with the most advanced disease burden, and potentially the most altered hepatic microenvironments, are paradoxically misclassified into the control group. Furthermore, the enzymatic components of the index are equally susceptible to distortion, while systemic inflammation can also suppress the hepatic synthesis of certain markers[11].

We acknowledge that the authors have performed a subgroup analysis using computed tomography to assess hepatic steatosis, which reinforces the robustness of their steatosis assessment[1]. However, it is important to note that conventional computed tomography, much like the HSI, primarily detects the density of macroscopic fat accumulation. It lacks the sensitivity to detect the subtle, necro-inflammatory changes or early-stage fibrosis that characterize the more aggressive forms of MASLD. Therefore, even with the addition of radiographic validation, the study fundamentally assesses the impact of hepatic fat quantity rather than hepatic inflammatory quality. Reliance on these steatosis-weighted measures in a cachectic population likely attenuates the true association toward the observation of no effect, leading to an underestimation of the prevalence and impact of MASLD. It is plausible that the group identified as free of MASLD in this study was inadvertently enriched with patients suffering from advanced cachexia, thus masking any survival disadvantage associated with MASLD.

To address this diagnostic gap, several more sensitive modalities have demonstrated clinical utility in characterizing the full spectrum of MASLD[12,13]. For example, vibration-controlled transient elastography (FibroScan) and magnetic resonance elastography (MRE) can offer superior diagnostic performance for the assessment of hepatic fibrosis. Magnetic resonance elastography, in particular, has demonstrated an area under the curve of 0.92-0.97 for staging liver fibrosis across multiple etiologies[14]. The integration of these advanced imaging modalities into future oncologic studies would enable a more precise phenotyping of the hepatic milieu, moving beyond the binary classification of steatosis to capture the inflammatory and fibrotic dimensions that are most relevant to metastatic colonization.

PHENOTYPIC HETEROGENEITY AND THE DILUTION OF HIGH-RISK SIGNALS

A second, critical conceptual limitation of the study is the treatment of MASLD as a homogeneous binary entity based purely on the presence or absence of steatosis. This approach overlooks the profound heterogeneity within the spectrum of MASLD and the distinct biological implications of its subtypes. Emerging evidence suggests that it is the inflammatory and fibrotic microenvironment, rather than simple hepatic steatosis, that actively promotes tumor progression and metastasis[15]. Separately, the severity of hepatic fibrosis has been identified as an independent determinant of long-term outcomes and mortality[16].

By pooling all patients with elevated scores or radiographic steatosis into a single category, the study inevitably mixes a large number of low-risk patients with simple steatosis together with a potentially smaller, high-risk subgroup harboring undiagnosed fibrosis or active inflammation. This phenotypic dilution dramatically reduces the statistical power to detect meaningful signals. Metabolic inflammation and fibrosis create a pro-tumorigenic extracellular matrix in the liver, characterized by the activation of hepatic stellate cells and the deposition of collagen and fibronectin[17]. This stiff microenvironment is biologically distinct from a merely fatty environment and is known to facilitate the trapping and colonization of circulating tumor cells[18]. The absence of a significant association in the study conducted by Chon et al[1] may simply reflect that fat accumulation is a poor surrogate for the specific inflammatory niche required for metastasis. If the study had stratified patients based on fibrosis markers, such as the fibrosis-4 index or imaging elastography, the results might have revealed a different correlation between the fibrotic subtype of MASLD and liver metastasis. The current analysis, by focusing on steatosis, likely obscures the contribution of the specific pathological features that actually promote tumor dissemination.

STATISTICAL BIASES: COMPETING RISKS AND REVERSE CAUSATION

From a statistical perspective, the aggressive nature of pancreatic cancer introduces a significant competing risk problem that conventional time-to-event analyses may not fully address. In the context of pancreatic cancer, where the median survival is notoriously short, death serves as a major competing event that precludes the observation of metachronous liver metastasis[19]. Patients with significant metabolic comorbidities, such as MASLD, diabetes, or cardiovascular disease, may have a higher hazard of dying from non-cancer causes or from local tumor progression and biliary sepsis before clinically detectable liver metastases have time to develop[20].

In standard survival analyses, these patients are often censored or treated in a manner that does not account for the fact that death prevented the outcome of interest. This can lead to a paradoxical statistical conclusion where healthier patients appear to have higher metastasis rates simply because they survive long enough to develop the metastasis. We acknowledge that the authors presented cumulative incidence curve[1]; however, it remains crucial to verify whether competing risk regression models were utilized to derive the hazard ratios, as standard Cox proportional hazards models can overestimate risk in the presence of high competing mortality[21]. The distinction is not merely academic: Simulation studies and empirical analyses in oncologic settings have demonstrated that the Fine-Gray model and cause-specific hazard models can yield substantially different estimates of cumulative incidence when the competing event rate is high[22]. For instance, a study comparing standard Kaplan-Meier estimates with competing risk analyses in cancers demonstrated that the Kaplan-Meier method overestimated the cumulative incidence of the event of interest by up to 10%-15% when competing mortality was substantial[23]. Without explicit adjustment for the competing risk of death, the incidence of liver metastasis in the MASLD group is likely underestimated. To mitigate this bias, future studies should employ the Fine-Gray sub distribution hazard model or cause-specific hazard models, report cumulative incidence functions for both the event of interest and the competing event, and perform sensitivity analyses comparing results across different analytical frameworks.

Besides, the issue of reverse causation remains formidable in advanced cancer. The tumor burden itself alters the metabolism of the host, leading to a browning of adipose tissue and increased lipolysis that can actively mobilize fat out of the liver[24,25]. Therefore, patients with the most aggressive biology may have less liver fat at the time of diagnosis, creating a spurious association where lower liver fat correlates with worse cancer outcomes. This lipodystrophy of cancer further complicates the interpretation of survival data, suggesting that the metabolic status of the liver at diagnosis may be more reflective of the catabolic drive of the tumor than the baseline metabolic health of the patient[26]. To address this confounding, future investigations should consider incorporating longitudinal assessments of hepatic steatosis, ideally comparing pre-cancer baseline imaging with imaging at the time of diagnosis, or utilizing Mendelian randomization approaches to infer causal relationships between genetically predicted MASLD and pancreatic cancer metastasis.

BIOLOGICAL PLAUSIBILITY: THE PRO-METASTATIC MICROENVIRONMENT

Finally, beyond the methodological limitations outlined above, a profound discordance remains between the clinical findings reported in this study and a robust body of mechanistic research supporting the concept of a pre-metastatic niche. It should be noted that a large number of researches have indicated that the liver is not only a passive recipient of metastasis, but also actively involved in the pathophysiological processes of the tumor growth[27-29]. MASLD provides a fertile soil for the dissemination of pancreatic cancer through several well-characterized molecular pathways that are inadequately captured by clinical surrogates. We believe it is crucial to delineate these mechanisms to highlight the biological plausibility that was overshadowed by the clinical definition of the disease (Figure 1).

Figure 1 Biological mechanisms of pancreatic cancer metastasis. A: In the inflammatory milieu, hepatic neutrophils undergo NETosis, expelling DNA webs decorated with granular proteins. These structures physically entrap circulating pancreatic cancer cells, acting as an initial scaffold for metastatic seeding; B: Activated hepatic stellate cells transdifferentiate to deposit a stiff extracellular matrix composed of collagen and fibronectin fibers. This rigid architecture facilitates cancer cell adhesion via integrin signaling and promotes survival; C: Lipid-engorged hepatocytes release exogenous fatty acids, which are actively taken up by cancer cells to fuel mitochondrial β-oxidation and support rapid biomass production; D: Tumor-associated macrophages adopt a protective phenotype, shielding the tumor from cytotoxic T-cells. Meanwhile, metabolically stressed senescent hepatocytes release senescence-associated secretory phenotype factors, which stimulate tumor proliferation in a paracrine manner.

A primary pathway through which MASLD influences liver metastasis is the modulation of hepatic inflammation and the formation of physical traps for circulating tumor cells[30]. It is well-established that metabolic dysfunction in the liver drives a chronic, low-grade inflammatory response through multiple interconnected pathways, including oxidative stress, insulin resistance, and gut microbiota dysbiosis[31,32]. In the context of pancreatic cancer, this inflammation exacerbates the metastatic potential. Specifically, a mechanism that warrants distinct attention is the formation of neutrophil extracellular traps[33-35]. In the inflammatory milieu of MASLD, hepatic neutrophils are primed to undergo a form of cell death known as NETosis, wherein they expel decondensed chromatin fibers decorated with granular proteins such as neutrophil elastase[36,37]. Recent evidence has elucidated that these DNA webs act as physical scaffolds within the hepatic sinusoids, effectively trapping circulating pancreatic cancer cells that would otherwise pass through the hepatic circulation[38,39]. The proteins associated with these traps can interact with the coiled-coil domain-containing protein 25 on the surface of cancer cells, activating the integrin-linked kinase-beta-parvin pathway, which enhances cell motility and proliferation[40,41]. By failing to account for the heightened baseline of neutrophil activation in steatohepatitis, clinical studies may overlook this critical adhesive interaction that initiates metastatic seeding.

Concurrently, the role of the liver in regulating immune responses is critical[42]. In MASLD, the chronic activation of the NOD-like receptor family, pyrin domain containing 3 inflammasome in hepatic cells produces interleukin-1 beta, which promotes the polarization of macrophages[43,44]. While early-stage inflammation involves pro-inflammatory phenotypes, the arrival of metastatic cells often educates the myeloid compartment toward an immunosuppressive phenotype[45]. These tumor-associated macrophages secrete transforming growth factor-beta, creating an immune-privileged niche that protects incipient metastatic foci from cytotoxic T-cell surveillance[46-48]. Moreover, the structural remodeling of the liver parenchyma in MASLD plays a decisive role in the metastasis of pancreatic cancer[49]. The transition from simple steatosis to steatohepatitis involves the activation of hepatic stellate cells, which transdifferentiate into myofibroblasts and deposit excessive amounts of extracellular matrix components, including fibronectin and collagen type I[50]. This stiffened matrix triggers integrin-dependent signaling pathways in cancer cells, enhancing their adhesion to the hepatic sinusoidal endothelium and promoting survival against anoikis[51].

Beyond the immune and structural landscape, the metabolic environment itself supports the aggressive behavior of cancer cells through specific lipid-dependent reprogramming[52]. The study by Chon et al[1] implies that hepatic fat is merely a bystander, yet biological data suggests it acts as a metabolic fuel source[53]. Pancreatic cancer cells exhibit remarkable metabolic plasticity[54]. Upon entering the lipid-rich environment of a steatotic liver, metastatic cells have been observed to upregulate the scavenger receptor[55]. This upregulation allows the cancer cells to avidly uptake exogenous fatty acids released by steatotic hepatocytes and adipocytes. These lipids are not merely stored but are utilized for mitochondrial beta-oxidation, providing the high energy yield required for rapid colonization and biomass production[56]. Additionally, the process of lipophagy allows cancer cells to degrade intracellular lipid droplets to sustain survival during nutrient stress[57]. This metabolic coupling between the steatotic host liver and the metastatic seed ensures that MASLD provides a distinct survival advantage to disseminated tumor cells, underscoring that the metabolic hepatic milieu is far from biologically inert in the context of metastatic colonization.

Furthermore, the concept of cellular senescence connects the metabolic toxicity of the liver to tumor progression[58]. Prolonged lipid accumulation leads to hepatocyte lipotoxicity and DNA damage, inducing a state of cellular senescence[59]. These senescent hepatocytes, while growth-arrested, remain metabolically active and develop a senescence-associated secretory phenotype. The secretome of these senescent cells is enriched with pro-inflammatory cytokines, growth factors such as amphiregulin, and matrix metalloproteinases[60]. This secretory cocktail acts in a paracrine manner to stimulate the proliferation of adjacent tumor cells and facilitate the remodeling of the vascular niche[61]. Thus, the aging and metabolically stressed liver becomes a proactive partner in the metastatic cascade.

CONCLUSION

The investigation conducted by Chon et al[1] provides valuable insights into the relationship between MASLD and the progression of pancreatic cancer. However, as discussed in detail, the reported result should be interpreted with significant caution, as the absence of a statistical correlation in a retrospective cohort does not equate to the absence of biological causation. Several methodological factors may have contributed to the observed result. First, the reliance on noninvasive surrogates such as the HSI, whose diagnostic accuracy is compromised in cachectic patients, likely introduced differential misclassification that attenuated the true prevalence and impact of MASLD. Second, the treatment of MASLD as a binary entity, without stratification by fibrotic or inflammatory subtypes, diluted the potential signal from the high-risk subgroup most relevant to metastatic colonization. Third, the statistical framework may not have fully accounted for the competing risk of death, which is exceptionally high in pancreatic cancer and can bias estimates of metastasis incidence. Only by integrating precise diagnostic phenotyping with mechanistic insights into the pre-metastatic niche can we fully elucidate the true impact of MASLD on pancreatic cancer progression.