Heat Shock Protein 47 as a Novel Predictive and Diagnostic Biomarker for Thrombosis in Hepatocellular Carcinoma

Introduction

Hepatocellular carcinoma (HCC) is the most common primary liver cancer and a leading cause of cancer-related mortality worldwide.¹ Its progression and patient survival are influenced by a multitude of factors, ranging from molecular and genetic alterations,^2–5 to critical changes in the tumor immune microenvironment.⁶ Beyond these intrinsic tumor biological drivers, clinical complications are pivotal determinants of outcomes.^7,8 Notably, patients with HCC face a substantially increased risk of thromboembolism (TE), particularly portal vein thrombosis (PVT), which occurs in 20–40% of cases.^9,10 The presence of PVT is a well-established marker of poor prognosis, significantly shortening overall survival^11,12. Cancer is a well-established risk factor for thrombosis, approximately 20% of all VTE are observed in patients with cancer, which confers a 9-fold higher risk of developing VTE.^7,13 Cancer-associated thrombosis involves a complex, multifactorial pathophysiological process, including cancer-related risk factors (such as hypercoagulability stemming from the procoagulant properties of cancer cells, cancer stage and site), patient-related risk factors (including gender and age), and other contributing factors (such as inflammation, chemotherapy, prolonged immobilization, and infection).^14–17

Most HCC cases arise in the context of liver cirrhosis (LC),¹⁸ which itself induces profound hemostatic alterations characterized by platelet activation, reduced synthesis of coagulation factors and anticoagulant proteins, and altered fibrinolytic activity.^19–21 These disturbances collectively shift the hemostatic balance toward a hypercoagulable state, elevating the risk of PVT and potentially accelerating HCC progression.^12,22 Early identification of prethrombotic states and timely prophylactic anticoagulation are therefore critical for improving clinical outcomes in these patients.

Current biomarkers for thrombosis, such as D-dimer, lack specificity in cancer patients due to frequent elevations in non-thrombotic conditions including inflammation, infection, and advanced age.²³ Thrombin generation assays (TGA), while informative, suffer from a lack of standardization, limiting their routine clinical use.²⁴ Thus, there is an urgent need for novel, specific biomarkers to predict thrombosis in high-risk populations like HCC.

Heat shock protein 47 (HSP47), encoded by the SerpinH1 gene, is a chaperone protein belonging to the serpin superfamily but lacks serine protease inhibitory activity.²⁵ As an endoplasmic reticulum-resident chaperone specific for collagen, HSP47 is essential for collagen biosynthesis and assembly.²⁶ It has been implicated in a range of pathological processes, most notably various fibrotic diseases and cancers, such as pulmonary/liver fibrosis, HCC, lung cancer and breast cancer, underscoring its role in fibrogenesis and tumor progression.^27–30In addition to these associations, emerging evidence highlights a direct role for HSP47 in promoting thrombus formation. HSP47 is specifically exposed on the surface of activated platelets, where it enhances glycoprotein VI (GPVI) binding to collagen, promoting platelet aggregation and coagulation, thereby promoting thrombus formation.^31,32 Evidence suggests that downregulation of HSP47 confers thromboprotective effects across diverse models, including long-term immobilized hibernating brown bears, mice, and spinal cord injury patients.³³ Furthermore, recent studies indicate that HSP47 inhibitors reduce cerebral vascular damage in ischemic stroke models.³⁴ This implies a dual role for HSP47 in both cancers and thrombosis. However, the relationship between HSP47 and thrombosis in HCC remains unexplored. More importantly, it is unknown whether HSP47 contributes to thrombosis via a unique, collagen-mediated pathway that is distinct from the coagulation cascades monitored by conventional biomarkers (eg, D-dimer). Therefore, to address these gaps, we investigated plasma HSP47 levels in HCC patients and evaluated its potential as a novel biomarker for predicting thrombotic events.

Methods

Patient Selection

This study enrolled 169 patients with HCC, 114 with LC, and 91 healthy controls (HC) from the First Affiliated Hospital of Nanchang University between January and March 2025. HCC diagnosis followed the Guideline for the Diagnosis and Treatment of Primary Liver Cancer (2024 Edition).³⁵ LC diagnosis was confirmed by clinical, laboratory, imaging, and histopathological data. TE diagnosis was performed using imaging. On ultrasonography, TE was identified as intraluminal material that was isoechoic or hypoechoic to soft tissue, causing partial or complete occlusion. On computed tomography, it was characterized by a filling defect that partially or totally occluded the vessel lumen.³⁶ The study was approved by the Hospital Ethics Committee (Approval No.: IIT2025 Clinical Ethics Review No. 224), with a waiver of informed consent, as it retrospectively analyzed only residual, anonymized serum samples from prior routine care, posing no additional risk to patients. All procedures complied with the Declaration of Helsinki.

Exclusion criteria included: intrahepatic cholangiocarcinoma, mixed HCC, extrahepatic malignancies, immune-related diseases, severe cardiopulmonary or renal dysfunction, systemic infections, recent anticoagulant/antiplatelet use, surgery within 30 days, blood product transfusion within 14 days, age < 18 years, pregnancy, or lactation. The HC group comprised 91 age- and sex-matched individuals without acute/chronic diseases or relevant medication use.

Data Collection

Clinical and laboratory data were retrieved from the hospital information system. Hematological parameters (WBC, RBC, HGB, PLT) were measured using a BC-6800PLUS automated hematology analyzer (Mindray, China). Coagulation parameters (PT, PT-INR, APTT, FIB, TT, D-dimer, FDP, PC, PS, AT-III) were analyzed using an ACL TOP 750 LAS automated coagulation analyzer (Werfen, Spain). Tumor characteristics (number, size, location, vascular invasion, metastasis) and CNLC stage were recorded for HCC patients.

Measurement of Plasma HSP47 Concentration

Plasma HSP47 levels were measured using a commercial Human Heat Shock Protein 47 (HSP47) ELISA Kit (Cat. No. RX105334H, Ruixin Biotechnology, Quanzhou, China). According to the manufacturer’s specifications, the kit has a detection limit of 0.1 pg/mL, with intra- and inter-assay coefficients of variation of < 10% and < 15%, respectively, and shows no cross-reactivity with structural analogues.

The assay was performed strictly according to the manufacturer’s protocol. Briefly, standards and samples were loaded in duplicate into a 96-well microplate. Subsequently, 100 μL of horseradish peroxidase (HRP)-conjugated antibody was added to each well and incubated at 37°C for 60 minutes. The plate was then washed five times with wash buffer. Next, 100 μL of prepared chromogenic reagent (a mixture of reagent A and reagent B) was added to each well. After gentle mixing, the plate was incubated at 37°C for 15 minutes in the dark. Finally, 50 μL of stop solution was added to each well. Optical density was measured at 450 nm within 15 minutes using a microplate reader (RT-6100, Rayto, China).

Statistical Analyses

In this retrospective study, sample size was based on available cases. To quantify the strength of the findings, post-hoc power was analyzed using G*Power 3.1 software based on HSP47 levels for five comparisons: TE (yes/no) and hypercoagulability (D-dimer > 0.5 mg/L; yes/no) within the HCC and LC groups, along with the comparison of CNLC stages (advanced vs. early) within the HCC group. The estimated power was 99%, 85%, 94%, 92% and 83%, respectively (two-tailed α = 0.05), underscoring the robustness of the findings.

Analyses were performed using R software (v4.2.1). Categorical variables were compared using Fisher’s exact, chi-square, or continuity-corrected chi-square tests. Continuous variables were tested for normality and homogeneity of variance; normally distributed data (mean ± SD) were analyzed using t-tests or ANOVA, while non-normal data (median, IQR) were analyzed using Mann–Whitney U or Kruskal–Wallis tests. Logistic regression identified thrombosis risk factors. ROC curve analysis evaluated the diagnostic performance of HSP47 and D-dimer. Two-tailed P < 0.05 was considered statistically significant.

Results

Patient Characteristics

The study included 169 HCC, 114 LC, and 91 HC participants. No significant differences were observed in age, sex, or BMI among groups (all P > 0.05). However, significant differences were found in alcohol consumption, smoking, diabetes, hypertension, and multiple laboratory parameters (WBC, RBC, HGB, PLT, PT, PT-INR, APTT, FIB, TT, D-dimer, FDP, PC, PS, AT-III; all P < 0.05). Compared to HC, both LC and HCC groups showed lower WBC, RBC, HGB, PLT, FIB, PC, PS, and AT-III (all P < 0.01), and higher PT, PT-INR, APTT, TT, D-dimer, and FDP (all P < 0.01). No significant difference in D-dimer or FDP was observed between HCC and LC groups (D-dimer, P = 0.395; FDP, P = 0.513) (Table 1).

Table 1 General Data Analysis of the HCC, LC and HC Groups

Plasma HSP47 levels were significantly elevated in the HCC group compared to both LC and HC groups (all P < 0.001; Figure 1A and B). The LC group also showed higher HSP47 levels than HC (P < 0.001; Figure 1C).

Figure 1 Comparison of plasma HSP47 levels among HCC, LC, and HC groups. HSP47, heat shock protein 47. (A) Comparison of plasma HSP47 levels between HCC and LC groups. (B) Comparison of plasma HSP47 levels between HCC and HC groups. (C) Comparison of plasma HSP47 levels between LC and HC groups. Bars indicated median and interquartile range (IQR). Pairwise comparisons were performed using the Mann–Whitney U-test. ***, P < 0.001.

Consistent with a role in tumor progression,³⁰ plasma HSP47 levels were significantly elevated in patients with larger tumors and in those with vascular invasion (all P < 0.01). In contrast, no significant differences were observed in relation to lymph node metastasis or distant metastasis (all P > 0.05). Plasma HSP47 levels increased significantly with advancing CNLC stage, (P for trend = 0.016) and were higher in advanced- versus early-stage disease (P = 0.004), indicating a clear association with disease progression (Table 2).

Table 2 Comparison of Plasma HSP47 Levels Among Different CNLC Stages in Patients with HCC

Comparison of Plasma HSP47 Levels Between Groups with High and Normal D-Dimer Levels in Patients with HCC and LC

Elevated D-dimer levels are associated with hyperfibrinolysis, which indicates a hypercoagulable state and increases the risk of thrombosis.²³ Using a cutoff level of D-dimer (0.5 mg/L),³⁷ we stratified HCC and LC patients into hypercoagulable and normal subgroups. In the HCC group, elevated D-dimer levels were significantly associated with WBC, RBC, HGB, PT, PT-INR, TT, FDP, PC, AT-III, and HSP47 (all P < 0.05). In the LC group, elevated D-dimer was associated with RBC, HGB, PT, PT-INR, FIB, TT, FDP, PC, AT-III, and HSP47 (all P < 0.05) (Table 3).

Table 3 Comparison of Laboratory Parameters Between Groups with High and Normal D-Dimer Levels in Patients with HCC and LC

In both HCC and LC groups, patients with elevated D-dimer exhibited significantly higher plasma HSP47 levels than those with normal D-dimer (all P < 0.01; Figure 2A–D). These results indicated that HSP47 elevation was associated with hypercoagulability across different etiologies.

Figure 2 Comparison of plasma HSP47 levels between groups with and without high D-dimer levels in patients with HCC and LC. (A) Comparison of plasma HSP47 levels between HCC patients with and without high D-dimer. (B) Comparison of plasma HSP47 levels between LC patients with and without high D-dimer. (C) Comparison of plasma HSP47 levels between HCC and LC patients with high D-dimer. (D) Comparison of plasma HSP47 levels between HCC and LC patients without high D-dimer. Bars indicated median and interquartile range (IQR). Pairwise comparisons were performed using the Mann–Whitney U-test. **, P < 0.01; ***, P < 0.001.

HSP47 Levels in Patients with Thrombosis

Patients were classified into four subgroups: HCC with TE, HCC without TE, LC with TE, and LC without TE. Median plasma HSP47 levels were significantly higher in patients with TE (HCC with TE: 904.7 [876.25–1016]; HCC without TE: 631.95 [299.6–816]; LC with TE: 596.03 [495.92–667.36]; LC without TE: 229.5 [108.53–630.76]).

In both HCC and LC groups, patients with TE exhibited significantly higher HSP47 levels than those without TE (P < 0.001; Figure 3A and B), with significant differences across all subgroups (P < 0.001; Figure 3C and D). These findings demonstrate a strong association between elevated HSP47 and thrombogenesis.

Figure 3 Comparison of plasma HSP47 levels between groups with and without TE in patients with HCC and LC. (A) Comparison of plasma HSP47 levels between HCC patients with and without TE. (B) Comparison of plasma HSP47 levels between LC patients with and without TE. (C) Comparison of plasma HSP47 levels between HCC and LC patients with TE. (D) Comparison of plasma HSP47 levels between HCC and LC patients without TE. Bars indicate median and interquartile range (IQR). Pairwise comparisons were performed using the Mann–Whitney U-test. ***, P < 0.001.

Risk Factors for Thrombosis in HCC

Univariate analysis identified LC, decreased RBC, prolonged PT-INR, decreased FIB, prolonged TT, increased D-dimer, increased FDP, decreased PC, decreased AT-III, and increased HSP47 as risk factors for TE (Table 4). Multicollinearity was absent (all VIF < 2.467). Multivariate logistic regression confirmed that elevated plasma HSP47 was an independent risk factor for TE in HCC (Table 4).

Table 4 Univariate and Multivariate Logistic Regression Analysis of TE Risk Factors in the HCC Patients

Diagnostic Performance of HSP47 and D-Dimer

ROC analysis for diagnosing thrombosis in HCC yielded an AUC of 0.859 (95% CI: 0.780–0.939) for HSP47, with 81.5% sensitivity and 82.4% specificity (Table 5). The AUC for D-dimer was 0.740 (95% CI: 0.640–0.840), with 74.1% sensitivity and 65.5% specificity (Table 5). HSP47 demonstrated significantly superior diagnostic performance compared to D-dimer (P < 0.05; Figure 4).

Table 5 ROC Curve Parameters for HSP47 and D-Dimer

Figure 4 ROC curves of HSP47 and D-dimer for the diagnosis of HCC-associated thrombosis, with the diagonal line indicating the reference line (AUC = 0.5). The black line represents HSP47, and the red line represents D-dimer.

Discussion

Patients with both LC and HCC are at high risk of TE, with PVT being the most prominent subtype.^9,10,38,39 This study established HSP47 as a novel, independent biomarker for predicting and diagnosing thrombosis in HCC. PVT was the predominant TE subtype, occurring in 85.2% of HCC and 89.5% of LC patients with TE, aligning with its known role as a critical prognostic factor.^11,12 Early PVT diagnosis is therefore essential for improving patient outcomes.

While D-dimer is a sensitive marker for venous thromboembolism,⁴⁰ its low specificity in cancer and other conditions limits its diagnostic utility.^41,42 Our results confirm that D-dimer, though a risk factor, is not an independent predictor of thrombosis in HCC, highlighting the need for better biomarkers.

HCC commonly arises in the setting of LC,¹⁸ In both conditions, the combined impact of the underlying liver pathology and its treatments (eg, transarterial chemoembolization, systemic therapy) compromises hepatic synthetic function.⁴³ Given that the liver produces most coagulation and anticoagulation factors, this impairment exacerbates dysregulation of coagulation and fibrinolysis and promotes aberrant platelet activation.^44,45 Since activated platelets express HSP47,³³ which further potentiates thrombus formation, a vicious cycle may be established. In our study, compared to the HC group, patients with LC or HCC showed elevated levels of PT, D-dimer, and FDP, along with reduced activities of FIB, PC, PS, and AT-III. These abnormalities were more marked in patients who developed thrombosis, collectively indicating a systemic hypercoagulable state in HCC.^46–49

HSP47 is a collagen-specific chaperone responsible for the correct folding and stabilization of procollagen in the endoplasmic reticulum, which is essential for maintaining the structural integrity of the extracellular matrix.^50,51 In HCC, tumor cells and surrounding stromal cells highly express HSP47, leading to abnormal perivascular deposition of type I collagen, exacerbating fibrosis and inducing pathological vascular remodeling.^52–54 This process not only promotes the growth, migration, and invasion of HCC but also results in thickening, stiffening, and increased vulnerability of the vessel wall.⁵³ Following vascular endothelial injury, HSP47-mediated collagen becomes extensively exposed, serving as a key trigger for platelet activation, aggregation, and thrombosis.^27,32 Under the stimulation of collagen, platelets undergo rapid aggregation and release a variety of procoagulant substances, which in turn activate the coagulation cascade and ultimately lead to thrombus formation.⁴⁵ Notably, activated platelets also express HSP47 on their surface, which enhances the binding of GPVI to collagen, further promoting platelet aggregation and coagulation, thereby forming a positive feedback loop that exacerbates thrombus formation.³³

In addition, HCC involves tumor-specific prothrombotic mechanisms.¹⁵ A persistent inflammatory response is observed throughout HCC progression,⁵⁵ fostering a chronic inflammatory milieu that promotes vascular endothelial injury, aberrant platelet activation, and enhanced neutrophil recruitment—all contributing to thrombus formation.⁵⁶ In this setting, HSP47 on the surface of activated platelets potentiates platelet-collagen interaction, hastening thrombus development.^31,32 Furthermore, tumor-derived small extracellular vesicles (EV) in HCC contribute to platelet hyperactivation, establishing a positive feedback loop that amplifies platelet aggregation and thereby increases thrombosis risk.^57,58 Emerging evidence shows that these EVs carry HSP47 on their surface, which may indirectly stimulate thrombosis via interactions with the matrix—consistent with the known role of platelet-bound HSP47 in enhancing GPVI signaling.⁵³ Within the tumor microenvironment, extensive neutrophil infiltration triggers the release of neutrophil extracellular traps (NETs), which in turn enhance local coagulation activity and facilitate thrombosis.^59,60 HSP47 has been shown to directly facilitate NETosis, thereby exacerbating thrombus propagation.³³ In HCC-associated thrombosis, HSP47 drives fibrosis, molds a pro-thrombotic matrix, and directly engages platelet activation cascades, thereby systematically coupling HCC progression with thrombotic mechanisms and establishing HSP47 as a key pro-thrombotic hub specific to the tumor microenvironment, distinct from conventional coagulation indicators.

We observed that plasma HSP47 levels were significantly elevated in the HCC cohort relative to the LC and HC groups. Moreover, HSP47 levels increased with advancing tumor stage and correlated with aggressive tumor features, supporting a link to disease progression. Notably, levels were even higher in HCC patients with hypercoagulability (D-dimer > 0.5 mg/L) or active thrombosis, indicating that thrombotic complications are associated with a notable additional rise in HSP47. While univariate analysis indicated HSP47 as a risk factor for thrombosis, multivariate analysis confirmed that it remained an independent predictor of thrombotic risk in HCC. The diagnostic utility of HSP47 was underscored by ROC curve analysis, which revealed a sensitivity of 81.5% and a specificity of 82.4%. These results strongly suggested that plasma HSP47 serves as a robust and independent diagnostic biomarker for thrombosis in the context of HCC.

However, several limitations of this study should be considered when interpreting the results. First, as a single-center study, our findings require validation in multi-center, geographically diverse populations to enhance their generalizability, additionally, the retrospective nature of the study limited the sample size to available cases (though post-hoc power for primary outcomes was >80%), highlighting the need for prospective validation with pre-planned enrollment. Second, the measurement of plasma HSP47 was conducted at a single time point; therefore, dynamic changes in its expression throughout disease progression or in response to therapy remain unexplored. Third, although we have strict inclusion standards, we cannot avoid the fact that HCC patients who are likely to have a group of patients have been treated with anti-tumor therapy (such as interventional therapy, chemotherapy, or immunotherapy), which cannot completely rule out the potential impact of these treatments on HSP47 levels. Fourth, our analysis of thrombus types was constrained by the natural distribution within our cohort. While PVT was predominant, the very limited number of non-PVT cases precluded a meaningful statistical comparison of HSP47 levels across different thrombus types, which represents an important clinical dimension for future investigation. Consequently, future studies with larger, prospective cohorts and serial biomarker assessments are warranted to definitively establish the clinical utility of HSP47 for the early diagnosis of thrombosis in HCC.

Conclusion

This study identified HSP47 as an independent risk factor and a superior diagnostic biomarker for thrombosis in HCC patients. Its strong predictive performance underscores its potential clinical utility for early risk stratification and timely intervention, paving the way for improved management of HCC-associated thrombosis. Moreover, HSP47 elevated in advanced HCC and further increased with thrombosis, may mechanistically connect tumor progression to thrombogenesis. This conceptual advance beyond conventional coagulation models requires direct testing. Consequently, future prospective, multicenter studies are needed to validate its prognostic utility and delineate its pathogenic role, paving the way for clinical translation.