PCSK6 and clinical outcomes in the Pulmonary Fibrosis Foundation Patient Registry

Abstract

Background

Higher plasma proprotein convertase subtilisin/kexin type 6 (PCSK6) concentration has been associated with increased all-cause mortality in idiopathic pulmonary fibrosis (IPF). Its role across the broader spectrum of fibrotic interstitial lung disease (ILDs) remains unclear. We examined associations between PCSK6 and respiratory outcomes in participants with pulmonary fibrosis in the Pulmonary Fibrosis Foundation Patient Registry and evaluated its expression in fibrotic lungs.

Methods

PCSK6 concentration was measured in plasma from 428 participants with fibrotic ILD using ELISA and tested for association with time to death, lung transplant, first respiratory hospitalization, and 10% decline in forced vital capacity (FVC) using adjusted Cox models. Single-cell RNA sequencing (scRNA-seq) data obtained from the IPF Cell Atlas were used to examine PCSK6 expression across disease states.

Results

PCSK6 levels were not significantly associated with time to death, lung transplantation, respiratory hospitalization, or 10% FVC decline in the overall cohort, or in the subgroup of participants with usual interstitial pneumonia pattern of fibrosis. Among participants with IPF, higher PCSK6 was associated with reduced survival, consistent with prior findings. scRNA-seq of fibrotic lungs demonstrated high PCSK6 expression in HAS1 + fibroblasts.

Conclusion

PCSK6 was not associated with worse respiratory outcomes in mixed fibrotic ILD but was associated with reduced survival in IPF. Its enrichment in HAS1 + fibroblasts suggests this pathway may be specific to IPF pathogenesis.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12931-026-03642-1.

Background

Idiopathic pulmonary fibrosis (IPF) is a progressive fibrosing interstitial lung disease (ILD) marked by declining lung function and high mortality. A subset of individuals with other fibrotic ILDs exhibit a similar clinical trajectory, termed progressive pulmonary fibrosis (PPF) [1]. It remains unclear whether IPF and PPF, particularly in the presence of a usual interstitial pneumonia (UIP) pattern, reflect distinct disease entities, or represent points along a shared pathobiological continuum with common fibrotic mechanisms that span disease categories [2–9].

Proprotein convertase subtilisin/kexin type 6 (PCSK6) is a calcium-dependent serine endoprotease that acts as a mediator of transforming growth factor-β signaling, and has been linked to increased collagen deposition and extracellular matrix formation [10–12]. PCSK6 is highly expressed in airway epithelium, adventitial fibroblasts, and lymphatic endothelial cells in the lung [13]. A recent study identified PCSK6 as a potential prognostic blood biomarker in three independent IPF cohorts. Higher circulating PCSK6 levels were associated with worse transplant-free survival but associations with other respiratory outcomes among patients with IPF and other fibrotic ILDs are unknown [13].

In this study, we aimed to investigate PCSK6 expression and its association with respiratory-related clinical outcomes in individuals with IPF and non-IPF fibrotic ILDs enrolled in the Pulmonary Fibrosis Foundation’s Patient Registry (PFF-PR). We hypothesized that higher blood PCSK6 levels are associated with worse respiratory outcomes in all fibrotic ILDs and that this association is stronger among those with a UIP pattern on chest imaging. Furthermore, we examined PCSK6 expression in fibrotic lungs using publicly available single-cell RNA sequencing (scRNA-seq) datasets.

Methods

Participants

The Pulmonary Fibrosis Foundation Patient Registry (PFF-PR) is a US-based cohort of 2003 participants aged 18 years or older with a diagnosis of fibrotic ILD who had not undergone lung transplant, were seen at registry centers, and provided informed consent starting in 2016 [14]. Participants provided a one-time blood sample at the time of Registry enrollment. Baseline forced vital capacity (FVC) and diffusion capacity of the lungs for carbon monoxide (DLCO) were obtained from pulmonary function tests performed clinically closest to date of enrollment. Patient data were abstracted every 6 months. We measured PCSK6 in plasma samples from 428 participants randomly selected from the Registry using commercially available ELISA kits (MyBioSource, San Diego, CA) followed by manufacturer’s protocol. PCSK6 measurements were truncated at lower and upper limits of detection (31 and 1000 ng/mL, respectively).

Image analysis

Chest CT scans of PFF-PR participants were performed for clinical purposes at participating sites. Baseline CT scans were defined as the CT study closest to the date of patient consent/enrollment meeting quality requirements for analysis [15]. Images missing inspiratory series, inadequate lung coverage, and/or severe motion or metal artifact were excluded [6]. Median time between CT scan and PCSK6 measurement in the overall cohort is -3.4 months (interquartile range −11.3 to 2.8 months). CT scans underwent standardized visual assessment for UIP classification at National Jewish Health [15]. Additionally, a deep learning algorithm developed using multiple instance learning (MIL) was used to objectively classify UIP. This algorithm has been demonstrated to be more sensitive than visual assessment in predicting histologic UIP [16]. Data-driven textural analysis (DTA) is a deep learning technique consisting of convolutional neural network algorithms used to quantify extent of fibrosis on CT images [17]. Subject-level DTA scores have previously been computed as the percentage of lung pixels classified as fibrotic on a subset of participants within the Registry who had technically adequate baseline CT scans [15].

PCSK6 cellular and tissue expression

We utilized scRNA-seq data from publicly available datasets to visualize the distribution of PCSK6 transcript expression across major lung cell populations. We leveraged the IPF Cell Atlas (http://www.ipfcellatlas.com/), which includes donor control (n = 10), IPF (n = 12) and other fibrotic ILDs (n = 8), including diagnoses of chronic hypersensitivity pneumonitis (n = 3), nonspecific interstitial pneumonia (n = 2), sarcoidosis (n = 2), and unclassifiable ILD (n = 1) [18–20]. We performed limited immunohistochemical (IHC) staining for PCSK6 using rabbit PCSK6 antibody (HPA004774, Atlas antibodies, Sweden) in archived fibrotic lung tissue obtained from the University of Virginia Biorepository and Tissue Research Facility to provide contextual visualization of protein distribution across ILD subtypes.

Statistical analysis

To assess the association between circulating PCSK6 levels and radiographic extent of lung fibrosis, a partial correlation analysis was performed between log2-transformed PCSK6 levels and [1] baseline DTA score [2], baseline FVC and [3] baseline DLCO, adjusting for age and sex as covariates. Partial correlation coefficients and p-values were computed using the pcor.test() function from the ppcor R package in RStudio 2024.12.0. We regressed log2-transformed PCSK6 levels and fibrosis score/FVC/DLCO separately on age and sex using linear models and extracted residuals. These residuals plotted against each other with a fitted linear regression line and 95% confidence interval.

Associations between log2-transformed plasma PCSK6 (ng/mL) and primary endpoints time to death (overall survival), lung transplant, first respiratory hospitalization, and 10% decline in relative FVC were based on a time-to-event analysis using Cox proportional hazard models. The proportional hazards assumption was assessed using Schoenfeld residuals and was found to be satisfied for covariates included in the model. Models were adjusted for age, sex, baseline FVC, baseline DLCO, and study site as a cluster effect. Missing covariate data were imputed using multiple imputation by chained equations. All analyses were first performed in the overall cohort. We then performed subgroup analyses based on ILD diagnosis, visual UIP classification, and MIL-UIP pattern on CT. MIL-UIP cutoff of 0.5 was used based on a previously published ROC analysis of MIL-UIP prediction of histologic UIP [16]. In secondary analyses, we truncated clinical follow-up to examine associations with 1-year outcomes. Hazard ratios were reported per doubling of PCSK6 concentration. Kaplan-Meier curves were stratified by tertile of PCSK6 with log-rank tests used to compare survival distributions. To formally assess potential effect modification, we added interaction terms between log₂(PCSK6) and sex, age category, MIL-UIP score, and CT pattern in multivariable Cox models for each outcome. Analyses were performed in the overall cohort and separately within IPF and non-IPF ILD cohorts. Because multiple interaction tests were conducted within each outcome, interaction p-values were adjusted using the Benjamini-Hochberg false discovery rate (FDR) method by outcome. Reported p-values reflect FDR adjustment.

Results

PCSK6 and disease severity

In the overall cohort, the median (IQR) age was 70 (12) years, 272 (64%) were male, and 275 (64%) had a diagnosis of IPF (Table 1). PCSK6 concentrations were similar between IPF and non-IPF patients. IPF patients were older, were more likely to be male and have UIP pattern on imaging, and a larger percentage experienced adverse clinical outcomes.

Table 1.

Baseline characteristics of PFF-PR participants with measured plasma concentrations of PCSK6 at time of enrollment

	Overall (n = 428)	IPF (n = 275)	Non-IPF fibrotic ILD (n = 153)		p-value
Age (years)	70 (63, 75)	72 (66, 77)	67 (59, 73)		< 0.001
Female	156 (36%)	66 (24%)	90 (59%)		< 0.001
Lung function*
Baseline FVC, liters	2.51 (2.0, 3.12)	2.58 (2.11, 3.16)	2.30 (1.80, 3.04)		0.012
Baseline FVC, % predicted	67 (56, 79)	66 (56, 78)	67 (55, 81)		0.6
Baseline DLCO, ml/min/mmHg	11.5 (9.0, 14.8)	11.5 (9.0, 14.9)	11.5 (9.0, 14.7)		0.8
Baseline DLCO, % predicted	41 (32, 52)	39 (31, 50)	44 (34, 54)		0.011
CT analysis*
MIL-UIP value	0.8 (0.59, 0.88)	0.85 (0.74, 0.90)	0.60 (0.16, 0.80)		< 0.001
Parenchymal Pattern					< 0.001
Definite or probable UIP	149 (52%)	122 (68%)	27 (25%)
Indeterminate for UIP	82 (29%)	43 (24%)	39 (36%)
Alternative pattern	56 (20%)	15 (8.3)	41 (38%)
DTA total fibrosis extent, %	27.68 (18.50, 42.62)	29.43 (22.06, 44.83)		23.16 (9.81, 39.86)
PCSK6 (ng/mL)	91 (60, 177)	89 (59, 155)	97 (61, 218)		0.14
Follow-up events
Follow-up time (days)	1,198 (535, 1,699)	1,106 (506, 1,645)	1,461 (646, 1,826)		0.003
Death	149 (35%)	110 (40%)	39 (25%)		0.003
Lung transplant	55 (13%)	44 (16%)	11 (7%)		0.009
Respiratory hospitalization	134 (31%)	93 (34%)	41 (27%)		0.13
10% FVC decline*	224 (62%)	157 (68%)	67 (52%)		0.003

Values presented as median (interquartile interval) or n (%)

*FVC missing for 20 participants at baseline and 67 at follow-up; FVC % predicted missing for 26 participants; DLCO missing for 40 participants; DLCO % predicted missing for 46 participants; MIL-UIP and DTA missing for 150 participants; parenchymal pattern missing for 141 participants

A statistically significant correlation was not observed between PCSK6 and baseline FVC, DLCO, or DTA in the overall cohort (Figures E1-E3). In contrast, within the subset of participants with IPF, a moderate and statistically significant positive correlation was observed between PCSK6 and DTA (Figure E3B; r = 0.247, p = 0.0014), indicating that higher plasma PCSK6 levels were associated with greater radiographic fibrotic extent.

PCSK6 and clinical outcomes

Among all participants with fibrotic ILD over 74 months of follow-up, higher PCSK6 levels were not significantly associated with overall survival, lung transplantation, first respiratory hospitalization, or 10% FVC decline (Fig. 1). In IPF, greater PCSK6 was significantly associated with worse overall survival (HR 1.17, 95%CI 1.00-1.36), but not with respiratory-specific outcomes. This effect was stronger when the observation period for overall survival was truncated at 1 year but was not significant (HR 1.33, 95%CI 0.99–1.78). In non-IPF fibrotic ILDs, higher PCSK6 was associated with significantly reduced hazard of 10% FVC decline (HR 0.80, 95%CI 0.66–0.99). Among individuals with visually-assessed definite/probable UIP or with MIL-UIP, PCSK6 levels were not significantly associated with any clinical outcomes (Figure E4).

Fig. 1.

Associations between circulating plasma protein concentration of PCSK6 and time to clinical outcome over total length of follow-up. Hazard ratios are reported per doubling of PCSK6. Bold confidence intervals indicate that the interval does not include 1.0

The direction of association between PCSK6 tertiles and survival differed by cohort (Figure E5). In IPF, individuals in the lowest PCSK6 tertile showed a trend toward better survival (log-rank p = 0.12). Conversely, in non-IPF fibrotic ILDs, the lowest tertile demonstrated a trend toward worse survival (log-rank p = 0.09).

To evaluate potential effect modification, interaction terms between log₂(PCSK6) and sex, age category, MIL-UIP score, and CT pattern were added to multivariable Cox models for each outcome (Tables E1–E4). After false discovery rate adjustment for multiple testing within outcomes, there was no consistent evidence that demographic or radiologic characteristics modified the association between PCSK6 and clinical outcomes. An interaction between PCSK6 and age was observed for mortality in the non-IPF ILD cohort (adjusted p = 0.037), and an interaction between PCSK6 and sex was observed for 10% FVC decline in the non-IPF ILD cohort (adjusted p = 0.002); however, no robust or consistent interaction patterns were identified across outcomes.

Cellular and tissue expression

Using publicly available scRNA-seq datasets from individuals with pulmonary fibrosis, we identified highest PCSK6 expression in lymphatic cells, followed by fibroblasts (Fig. 2A-C). Figure 2E illustrates that PCSK6 is predominantly expressed in IPF compared to other ILD subtype or control. In IPF lungs, PCSK6 was primarily expressed in HAS1 + fibroblasts compared to other fibroblast subtypes, which are uniquely expressed in IPF lungs and not other fibrotic ILDs (Fig. 2D). In contrast, PCSK9 was mainly expressed in epithelial cells (Fig. 2D), implying cell type specific function of PCSK isoforms.

Fig. 2.

Cell type specific expression of PCSK6 in IPF lung. A Feature plot of PCSK6 expression in IPF scRNA-seq data. B Bar plot of PCSK6 expression in airway and lung parenchyma using IPF scRNA-seq data. C Dot-plot demonstrating greater PCSK6 expression in HAS1 + fibroblast cells versus non-HAS1 + fibroblasts. In contrast, PCSK9 was predominantly expressed in epithelial cells. The expression intensity is indicated by the color purple, and the size of the cell population expressing the gene is indicated by the circle size as noted in the legend. D Dot-plot demonstrating greater PCSK6 expression in IPF compared to control and other ILDs, including non-specific interstitial pneumonia (NSIP) and chronic hypersensitivity pneumonitis (cHP)

IHC staining revealed PCSK6 was present in the ciliated epithelial and alveolar epithelial cells in areas of fibrosis in all six patients (Figure E6 – PCSK6 stained in orange-brown). The mean PCSK6 value as measured by staining intensity was higher in IPF (4.09%) as compared to FHP (1.72%) and NSIP (1.67%), though the difference did not reach statistical significance (p = 0.11).

Discussion

In a multi-center cohort of patients with fibrotic ILDs, we confirmed prior observations that higher plasma PCSK6 levels are associated with worse survival in IPF [13]. In contrast to our initial hypothesis, we did not observe similar associations in non-IPF fibrotic ILDs, including among those with a UIP pattern on imaging. These findings suggest that PCSK6 may reflect biologic processes which are unique to IPF rather than a unifying biomarker across fibrotic ILD subtypes. Furthermore, these results challenge the concept that UIP represents a single biological entity across ILD subtypes, and suggest that radiologic UIP does not fully capture mechanistic differences which differentiate IPF from other fibrotic ILDs [21].

Interestingly, higher PCSK6 levels were associated with a lower risk of 10% FVC decline in non-IPF ILDs. This finding is counterintuitive given the association between PCSK6 and increased mortality in the IPF cohort, and could reflect divergent biologic roles of PCSK6 across ILD subtypes. This observed association may alternatively reflect confounding by indication, whereby patients with higher PCSK6 levels exhibited disease features prompting earlier or more intensive therapy, which in turn attenuated subsequent lung function decline. Because treatment approaches differ between IPF and non-IPF ILDs, differences in associations between PCSK6 and clinical outcomes may partly reflect treatment effects rather than intrinsic disease biology. In addition, longitudinal FVC analyses may be susceptible to competing-risk and survivor biases, which may preclude observation of FVC decline.

In further support of an IPF-specific role of PCSK6, we observed that plasma PCSK6 levels correlated with greater radiological quantitative fibrosis extent in IPF but not in other ILDs. scRNA-seq analysis also demonstrated greater PCSK6 expression in IPF compared to other ILDs, with complementary IHC findings. Within IPF lungs, greatest PCSK6 expression was observed in HAS1 + lung fibroblasts compared to other fibroblast subtypes, a cell population that is unique to IPF [20, 22].

The connection between PCSK6 and mortality in IPF may reflect systemic pathways beyond the lung. Dysregulated PCSK6 expression has been linked to cardiovascular disease and cardiac remodeling following myocardial ischemia [11, 23–25]. Individuals with IPF have more than a 3-fold higher risk of acute coronary syndrome compared to the general population, with cardiovascular disease (CVD) being the second leading cause of death in IPF [26, 27]. CVD is significantly more prevalent in IPF than in COPD, despite similar underlying risk factors and patient characteristics [28–30]. PCSK6 is the most enriched PCSK in atherosclerotic plaques, where it is highly expressed in proliferating smooth muscle cells, inducing smooth muscle cell migration in response to platelet-derived growth factor subunit B and playing a key role in vascular remodeling of atherosclerosis [25]. Higher blood PCSK6 levels are positively associated with adverse CVD events [31]. Our findings may reflect unique mechanisms connecting IPF and CVD conditions, and highlight the need to better address cardiovascular disease as an important contributor to mortality in IPF. The absence of associations with respiratory-specific outcomes supports this interpretation. Incorporating comorbid conditions into prognostic frameworks and investigating the broader biologic roles of PCSK6 in IPF may improve risk stratification and clinical decision making in this patient population.

In participants with IPF, we observed that the association between PCSK6 and all-cause mortality is stronger at one year and attenuates over longer follow up, demonstrating that a single timepoint biomarker measurement may offer short-term prognostic value but may be less informative over an extended disease course. This observation highlights the potential benefit of incorporating serial biomarker assessments into prognostic frameworks.

Despite the strength of a large, multicenter cohort and integration of both quantitative imaging and scRNA-seq data, several limitations should be acknowledged. IHC was performed on a limited number of tissue samples from a single center, without healthy control lungs and without cell-type-specific co-staining. Clinical diagnoses of tissue specimens predated the most recently published diagnostic guidelines, which may introduce misclassification. Time-to-event analysis was limited by local practice variation in diagnosis and treatment strategies, and varying stages of disease severity at time of enrollment [14]. For imaging analyses, there is variability in time between baseline CT acquisition date and date of blood draw used to measure PCSK6 level amongst participants. Finally, the absence of association with respiratory-specific outcomes suggests that further investigation is needed to investigate extrapulmonary processes involving PCSK6 in patients with IPF; unfortunately data regarding cause of death including respiratory, cardiovascular, or alternate is not available in the PFF-PR cohort. These findings require validation in independent cohorts with harmonized biomarker and imaging data before firm conclusions regarding disease-specific prognostic utility can be drawn.

Conclusions

Higher PCSK6 levels are associated with worse overall survival in IPF, but not with other respiratory outcomes or in non-IPF fibrotic ILDs. PCSK6 is preferentially expressed by HAS1 + fibroblasts which are unique to IPF, and plasma levels are correlated with quantitative extent of fibrosis in these patients.

Abbreviations

IPF

Idiopathic pulmonary fibrosis

ILD

Interstitial lung disease

PPF

Progressive pulmonary fibrosis

UIP

Usual interstitial pneumonia

(PCSK6) or 9 (PCSK9)

Proprotein convertase subtilisin/kexin type 6

PFF-PR

Pulmonary Fibrosis Foundation’s Patient Registry

scRNA-seq

Single-cell RNA sequencing

MIL

Multiple instance learning

DTA

Data-driven textural analysis

IHC

Immunohistochemical

FHP

Fibrotic hypersensitivity pneumonitis

NSIP

Nonspecific interstitial pneumonia

FVC

Forced vital capacity

DLCO

Diffusion capacity of the lungs for carbon monoxide

IQR

Interquartile range

Authors’ contributions

KB and SK prepared the manuscript text. KB, JO, JK and AP were involved in conceptual study design. KB, WW, AP and WS performed statistical analysis of PFF-PR data and prepared associated figures. SK, CJ, CM, and JK performed IHC analysis and prepared the associated figure. SH and DL performed CT image analysis. SM, YH, IN, and JK performed sc-RNA analysis and prepared associated figures. All authors reviewed the manuscript.

Funding

Supported by the National Institutes of Health National Heart, Lung, and Blood Institute grants K23 HL140199 (AJP), K23 HL150301 (JSK), and T32 HL134629 (KNB).

Data availability

The datasets analyzed during the current study are available from the Pulmonary Fibrosis Foundation Patient Registry, and from IPF Cell Atlas (http://www.ipfcellatlas.com/).

Declarations

Ethics approval and consent to participate

The study was conducted in accordance with approvals obtained following IRB review at Weill Cornell Medical Center and the University of Virginia Medical Center.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.