Association of Progressive PPFE and Mortality in Lung Cancer Screening Cohorts

Shahab Aslani; Mehran Azimbagirad; Daryl Cheng; Daisuke Yamada; Ryoko Egashira; Adam Szmul; Justine Chan-Fook; Robert Chapman

q-bio.QMMar 10, 2026

Association of Progressive PPFE and Mortality in Lung Cancer Screening Cohorts

Authors: Shahab Aslani, Mehran Azimbagirad, Daryl Cheng, Daisuke Yamada, Ryoko Egashira, Adam Szmul, Justine Chan-Fook, Robert Chapman,

Organizations: Satsuma Lab, Hawkes Institute, University College London, London, UK · Department of Respiratory Medicine, University College London, London, UK · Department of Computer Science, University College London, London, UK · Guy’s Cancer Centre, Guy’s and St Thomas’ NHS Foundation Trust, London, UK · Radiology Department, Hospital de la Santa Creu i Sant Pau, Sant Quintí, Barcelona, Spain · Department of Medicine, University of Iceland, Iceland · NIHR Manchester Biomedical Research Centre, Manchester Academic Health Science Centre, Central Manchester University Hospitals, NHS Foundation Trust, Manchester, UK · Division of Informatics, Imaging and Data Sciences, School of Health Sciences, University of Manchester, Manchester, UK · UCL Hawkes Institute, University College London, London, UK · Lungs for Living, Department of Respiratory Medicine, University College London, London, UK

Abstract

Background: Pleuroparenchymal fibroelastosis (PPFE) is an upper lobe predominant fibrotic lung abnormality associated with increased mortality in established interstitial lung disease. However, the clinical significance of radiologic PPFE progression in lung cancer screening (LCS) populations remains unclear. Methods: We analysed longitudinal low-dose CT scans and clinical data from two LCS studies: National Lung Screening Trial (NLST; n=7,980); SUMMIT study (n=8,561). An automated algorithm quantified PPFE volume on baseline and follow-up scans. Annualised change in PPFE was derived and dichotomised using a distribution-based threshold to define progressive PPFE. Associations between progressive PPFE and mortality were evaluated using Cox proportional hazards models adjusted for demographic and clinical variables. In SUMMIT cohort, associations between progressive PPFE and clinical outcomes were assessed using incidence rate ratios (IRR) and odds ratios (OR). Findings: Progressive PPFE independently associated with mortality in both LCS cohorts (NLST: Hazard Ratio (HR)=1.25, 95% Confidence Interval (CI): 1.01--1.56, p=0.042; SUMMIT: HR=3.14, 95% CI: 1.66--5.97, p<0.001). Within SUMMIT, progressive PPFE was strongly associated with higher respiratory admissions (IRR=2.79, p<0.001), increased antibiotic and steroid use (IRR=1.55, p=0.010), and showed a trend towards higher modified medical research council scores (OR=1.40, p=0.055). Interpretation: Radiologic PPFE progression independently associates with mortality across two large LCS cohorts, and associates with adverse clinical outcomes. Quantitative assessment of PPFE progression may provide a clinically relevant imaging biomarker to identify individuals at increased risk of respiratory morbidity within LCS programmes.

Figures & tables

Figure 1: Flowchart illustrating the selection of participants from the NLST and SUMMIT datasets. From the NLST cohort (n = 15,000; 44,722 CT scans), scans reconstructed using both lung and soft-tissue kernels were retained, followed by the selection of participants with baseline and one-year follow-up CTs and available survival data, yielding a final cohort of 7,980 participants (15,960 CT scans). From the SUMMIT cohort (n = 13,035; 33,093 CT scans), participants with baseline and two-year follow-up CTs and available follow-up information were included. After excluding individuals without ILA data, the final SUMMIT cohort comprised 8,561 participants (17,122 CT scans).

Figure 2: Kaplan–Meier survival by longitudinal PPFE progression. NLST (ten-year follow-up) and SUMMIT (two-year follow-up) cohorts comparing overall survival between participants with progressive and non-progressive PPFE change. Shaded areas indicate 95% confidence intervals and tick marks denote censoring. Numbers at risk are displayed below each plot. Participants with progressive PPFE exhibited significantly reduced survival in both cohorts (log-rank p $<$ 0.001 for both NLST and SUMMIT), consistent with multivariable Cox models (Table 3 ) in which progressive PPFE remained independently associated with mortality.

Variable	NLST			SUMMIT
	Non-Progressive	Progressive*	p-value	Non-Progressive	Progressive*	p-value
Num Patients, (%)	7549 (94.6)	431 (5.4)	–	8437 (98.5)	124 (1.5)	–
Age y, median (range)	61 (55–74)	63 (54–74)	$\raisebox{0.45206pt}{$ < $}0.001$	65 (55–80)	66 (55–77)	0.104
Sex (F/M%)	42/58	36/64	0.020	42/58	31/69	0.021
Smoking py, median (range)	48 (30–272)	51 (21–162)	0.007	40 (2.5–400)	43 (16.5–310)	0.226
FVC %pred, mean $\pm$ SD	–	–	–	88.12 $\pm$ 16.51	84.92 $\pm$ 18.13	0.053

Table 1: Patient demographics, and clinical variables including age, sex, height, smoking history (pack-years), FVC, visual scores of ILA, and computer-based scores of PPFE in NLST and SUMMIT cohorts. PPFE (baseline): upper-lung pleuroparenchymal fibroelastosis quantified on the first scan, PPFE (follow-up): upper-lung pleuroparenchymal fibroelastosis quantified on the follow-up scan,

\Delta

PPFE: annualised change in computerised upper-lung PPFE between scans, ILA: visual scores of interstitial lung abnormality, FVC: forced vital capacity, Progressive:

\Delta

PPFE

\geq

0.41 cm 3 /year. PPFE (baseline), PPFE (follow-up),

\Delta

PPFE, Height, and FVC are described as Mean±SD. Age and smoking are described as medians (range). Sex is described as Female/Male %. ILA is described as number/%.

Variable	Hazard ratio	95% confidence interval	p-value	C-index
NLST
Age (y)	1.10	1.09–1.11	$\raisebox{0.45206pt}{$ < $}0.001$	0.63
Sex (male)	1.49	1.34–1.66	$\raisebox{0.45206pt}{$ < $}0.001$	0.54
Smoking (py)	1.01	1.01–1.01	$\raisebox{0.45206pt}{$ < $}0.001$	0.60
Height (cm)	1.01	1.00–1.01	$\raisebox{0.45206pt}{$ < $}0.001$	0.54
Baseline PPFE	1.06	1.01–1.12	0.015	0.52

Table 2: Association of progressive PPFE with mortality in univariable Cox proportional hazards models in NLST and SUMMIT cohorts. Univariable models fitted separately for each cohort. Models adjusted for a single covariate at a time to assess its univariate association with mortality. C-index values represent model discrimination for each variable. HRs in bold indicate p < 0.05. Baseline PPFE: upper-lung pleuroparenchymal fibroelastosis quantified on the first scan, Progressive PPFE:

\Delta

PPFE

\geq

0.41 cm 3 /year, ILA: visual scores of interstitial lung abnormality, FVC: forced vital capacity.

Variable	Hazard ratio	95% confidence interval	p-value	Model C-index
NLST
Age (y)	1.10	1.09–1.11	$\raisebox{0.45206pt}{$ < $}0.001$	0.66
Sex (male)	1.30	1.11–1.51	$\raisebox{0.45206pt}{$ < $}0.001$
Smoking (py)	1.01	1.01–1.01	$\raisebox{0.45206pt}{$ < $}0.001$
Height (cm)	1.00	0.99–1.01	0.925
Baseline PPFE	1.02	0.96–1.08	0.552

Table 3: Association of progressive PPFE with mortality in multivariable Cox proportional hazards models in NLST and SUMMIT cohorts. Models adjusted for age, sex, smoking (pack-years), height, baseline PPFE, and the interaction between baseline PPFE and progressive PPFE. C-index shown per cohort for the multivariable model. HRs in bold indicate p < 0.05. Baseline PPFE: upper-lung pleuroparenchymal fibroelastosis quantified on the first scan, Progressive PPFE:

\Delta

PPFE

\geq

0.41 cm 3 /year.

Variable	Hazard ratio	95% confidence interval	p-value	C-index
SUMMIT
Age (y)	1.08	1.06–1.10	$\raisebox{0.45206pt}{$ < $}0.001$	0.70
Sex (male)	1.45	1.11–1.89	0.005
Smoking (py)	1.00	1.00–1.01	0.051
ILA (present vs absent)	1.42	1.13–1.79	0.002
Height (cm)	1.00	0.98–1.01	0.431

Table 4: Multivariable Cox proportional hazards model assessing progressive PPFE and mortality in the SUMMIT cohort. Model adjusted for age, sex, smoking (pack-years), height, FVC % predicted, ILA, baseline PPFE, and the interaction between baseline PPFE and progressive PPFE. C-index shown for the multivariable model. HRs in bold indicate p < 0.05. Baseline PPFE: upper-lung pleuroparenchymal fibroelastosis quantified on the first scan, Progressive PPFE:

\Delta

PPFE

\geq

0.41 cm 3 /year, ILA: visual scores of interstitial lung abnormality, FVC: forced vital capacity.

Variable	Incidence Rate Ratio	95% confidence interval	p-value
Age (y)	1.051	1.033–1.070	$\raisebox{0.45206pt}{$ < $}0.001$
Sex (male)	0.696	0.563–0.860	$\raisebox{0.45206pt}{$ < $}0.001$
Smoking (py)	1.005	1.002–1.009	$\raisebox{0.45206pt}{$ < $}0.001$
FVC % predicted	0.967	0.962–0.973	$\raisebox{0.45206pt}{$ < $}0.001$
ILA (present vs absent)	1.275	0.979–1.661	0.070
Baseline PPFE	1.135	1.059–1.217	$\raisebox{0.45206pt}{$ < $}0.001$

Table 5: Negative Binomial GLM model for respiratory admissions in the SUMMIT cohort. Model adjusted for age, sex, smoking (pack-years), FVC % predicted, ILA, and baseline PPFE. IRRs in bold indicate p < 0.05. Baseline PPFE: upper-lung pleuroparenchymal fibroelastosis quantified on the first scan, Progressive PPFE:

\Delta

PPFE

\geq

0.41 cm 3 /year, ILA: visual scores of interstitial lung abnormality, FVC: forced vital capacity.

Variable	Odds Ratio	95% confidence interval	p-value
Age (y)	1.021	1.014–1.028	$\raisebox{0.45206pt}{$ < $}0.001$
Sex (male)	0.554	0.509–0.601	$\raisebox{0.45206pt}{$ < $}0.001$
Smoking (pack-years)	1.012	1.010–1.013	$\raisebox{0.45206pt}{$ < $}0.001$
FVC % predicted	0.972	0.970–0.975	$\raisebox{0.45206pt}{$ < $}0.001$
ILA (present vs absent)	1.126	1.134–1.417	$\raisebox{0.45206pt}{$ < $}0.001$
Baseline PPFE	0.949	0.913–0.986	0.007

Table 6: Ordinal logistic regression model for mMRC score in the SUMMIT cohort. Model adjusted for age, sex, smoking (pack-years), FVC % predicted, ILA, and baseline PPFE. Odds in bold indicate p < 0.05. Baseline PPFE: upper-lung pleuroparenchymal fibroelastosis quantified on the first scan, Progressive PPFE:

\Delta

PPFE

\geq

0.41 cm 3 /year, ILA: visual scores of interstitial lung abnormality, FVC: forced vital capacity.

Variable	Incidence Rate Ratio	95% confidence interval	p-value
Age (y)	0.994	0.986–1.002	0.201
Sex (male)	0.540	0.490–0.595	$\raisebox{0.45206pt}{$ < $}0.001$
Smoking (pack-years)	1.007	1.005–1.009	$\raisebox{0.45206pt}{$ < $}0.001$
FVC % predicted	0.982	0.979–0.985	$\raisebox{0.45206pt}{$ < $}0.001$
ILA (present vs absent)	1.009	0.881–1.155	0.891
Baseline PPFE	1.059	1.016–1.104	0.006

Table 7: Negative Binomial GLM model for steroid/antibiotic inflammatory medication use in the SUMMIT cohort. Model adjusted for age, sex, smoking (pack-years), FVC % predicted, ILA, and baseline PPFE. IRRs in bold indicate p < 0.05. Baseline PPFE: upper-lung pleuroparenchymal fibroelastosis quantified on the first scan, Progressive PPFE:

\Delta

PPFE

\geq

0.41 cm 3 /year, ILA: visual scores of interstitial lung abnormality, FVC: forced vital capacity.

Figure 3: Representative examples of automated PPFE quantification on axial CT images from the NLST and SUMMIT cohorts. The first column shows the original CT images, and the second column shows the corresponding CT images with automated PPFE segmentation overlaid. Highlighted regions (red) indicate PPFE identified by the automated segmentation model in the upper lung zones. The first row shows images from a 65-year-old female participant in the NLST cohort with a quantified PPFE volume of 8.85 cm 3 . The second row shows images from a 55-year-old male participant in the SUMMIT cohort with a quantified PPFE volume of 5.32 cm 3 .

Figure 4: Flowchart illustrating identification of baseline-only participants in SUMMIT cohort. This diagram outlines the selection of SUMMIT participants who did not undergo a follow-up CT scan. Starting from the full SUMMIT cohort (n=13,035), individuals with available baseline CT imaging were identified, after which participants with follow-up CTs were excluded. Additional exclusions were applied based on imaging quality and availability of clinical data. The final baseline-only subgroup consisted of 2,580 participants.

Variable	NLST	SUMMIT
Num Patients, (%)	7980 (100)	8561 (100)
Age y, median (range)	61 (54–74)	65 (55–88)
Sex (F/M%)	41/59	42/58
Smoking py, median (range)	48 (21–272)	40 (2.5–400)
FVC %pred, mean $\pm$ SD	–	88.07 $\pm$ 16.53
Height cm, mean $\pm$ SD	172.44 $\pm$ 9.98	168.93 $\pm$ 9.79

Table 8: Overall cohort characteristics in NLST and SUMMIT. Variables include age, sex, height, smoking history (pack-years), FVC, visual scores of ILA, and computer-based scores of PPFE. PPFE (baseline) = upper-lung pleuroparenchymal fibroelastosis quantified on the first scan, PPFE (follow-up) = upper-lung pleuroparenchymal fibroelastosis quantified on the follow-up scan, ILA = visual scores of interstitial lung abnormality, FVC = forced vital capacity. PPFE (baseline), PPFE (follow-up), Height, and FVC are described as Mean

\pm

SD. Age and smoking are described as median (range). Sex is described as Female/Male %. ILA is described as number/%.

Variable	Non-clinically Important PPFE	Clinically Important PPFE*	P-value
Num Patients	1733	847	–
Age y, median (range)	65 (55–78)	67 (55–78)	$\raisebox{0.45206pt}{$ < $}0.001$
Sex, F/M%	42/58	42/58	0.041
Smoking py, median (range)	65 (55–78)	43 (8–252)	0.024
FVC %pred, mean $\pm$ SD	84.35 $\pm$ 17.12	85.18 $\pm$ 16.79	0.242
Height cm, mean $\pm$ SD	167.68 $\pm$ 9.09	169.82 $\pm$ 10.53	$\raisebox{0.45206pt}{$ < $}0.001$

Table 9: Patient demographics and clinical variables in the SUMMIT cohort comparing clinically important PPFE (C-PPFE) vs non-clinically important PPFE (NC-PPFE). Baseline PPFE refers to upper-lung pleuroparenchymal fibroelastosis quantified on the baseline scan. Baseline PPFE, Height, and FVC are reported as mean

\pm

SD. Age and smoking are medians (range). Sex is Female/Male%. ILA is n(%).

Variable	Incidence Rate Ratio	95% confidence interval	p-value
Age (y)	1.060	1.042–1.078	$\raisebox{0.45206pt}{$ < $}0.001$
Sex (male)	0.847	0.686–1.045	0.121
Smoking (pack-years)	1.005	1.002–1.009	$\raisebox{0.45206pt}{$ < $}0.001$
FVC % predicted	0.976	0.970–0.982	$\raisebox{0.45206pt}{$ < $}0.001$
ILA (present vs absent)	1.546	1.208–1.980	$\raisebox{0.45206pt}{$ < $}0.001$
Baseline PPFE	1.238	1.167–1.314	$\raisebox{0.45206pt}{$ < $}0.001$

Table 10: Negative Binomial GLM model for respiratory admissions in the SUMMIT cohort. Model adjusted for age, sex, smoking (pack-years), FVC % predicted, and ILA. IRRs in bold indicate p < 0.05. Baseline PPFE: upper-lung pleuroparenchymal fibroelastosis quantified on the first scan, ILA: visual scores of interstitial lung abnormality, FVC: forced vital capacity.

Variable	Odds Ratio	95% confidence interval	p-value
Age (y)	1.019	1.007–1.031	0.002
Sex (male)	0.649	0.559–0.753	$\raisebox{0.45206pt}{$ < $}0.001$
Smoking (py)	1.011	1.008–1.015	$\raisebox{0.45206pt}{$ < $}0.001$
FVC % predicted	0.973	0.969–0.977	$\raisebox{0.45206pt}{$ < $}0.001$
ILA (present vs absent)	1.084	0.891–1.319	0.419
Baseline PPFE	1.012	0.951–1.077	0.701

Table 11: Ordinal logistic regression model for mMRC score in the baseline-only SUMMIT cohort. Model adjusted for age, sex, smoking (pack-years), FVC % predicted, and ILA. Odds in bold indicate p < 0.05. Baseline PPFE: upper-lung pleuroparenchymal fibroelastosis quantified on the first scan, ILA: visual scores of interstitial lung abnormality, FVC: forced vital capacity.

Variable	IRR	95% confidence interval	p-value
Age (y)	1.024	1.006–1.042	0.007
Sex (male)	0.874	0.703–1.088	0.229
Smoking (py)	0.995	0.990–1.001	0.085
FVC % predicted	0.987	0.981–0.993	$\raisebox{0.45206pt}{$ < $}0.001$
ILA (present vs absent)	0.848	0.617–1.166	0.312
Baseline PPFE	0.963	0.878–1.057	0.436

Table 12: Negative Binomial GLM model for steroid/antibiotic inflammatory medication use in the baseline-only SUMMIT cohort. Model adjusted for age, sex, smoking (pack-years), FVC % predicted, and ILA. IRRs in bold indicate p < 0.05. Baseline PPFE: upper-lung pleuroparenchymal fibroelastosis quantified on the first scan, ILA: visual scores of interstitial lung abnormality, FVC: forced vital capacity.

Mortality Prediction

Mortality prediction estimates the risk, timing, or cause of death from clinical records, images, signals, and population data.

See all 30 papers

Time-Conditioned and Multi-Time Survival Prediction from 2D PET/CT Projections in Lung Cancer
Jun 10, 2026Ashish Chauhan, Sambit Tarai, Elin Lundström +3Cancer Survival Prediction Survival Analysis
Radiology, Department of Surgical Sciences, Uppsala University, Uppsala, Sweden · National Academic Infrastructure for Supercomputing (NAISS), LinköpingJun University, Linköping, Sweden · Antaros Medical, Mölndal, Sweden +1
Structured Proxy Features for Multimodal NSCLC Survival Prediction from Pretreatment CT
Aug 1, 2026Huu Phong Nguyen, Delower Hossain, Ehsan Saghapour +2Cancer Survival Prediction Oncology
Department of Biomedical Informatics and Data Science, School of Medicine, The University of Alabama at Birmingham (UAB), Birmingham, AL, United States. · Department of Computer Science, The University of Alabama at Birmingham (UAB), Birmingham, AL, United States. · System Pharmacology and AI Research Center (SPARC), The University of Alabama at Birmingham (UAB), Birmingham, AL, United States.
Time-driven Survival Analysis from FDG-PET/CT in Non-Small Cell Lung Cancer
Apr 8, 2026Sambit Tarai, Ashish Chauhan, Elin Lundström +5Cancer Survival Prediction Survival Analysis
Radiology, Department of Surgical Sciences, Uppsala University, Uppsala, SE-75185, Sweden · Molecular Imaging and Medical Physics, Department of Surgical Sciences, Uppsala University, Uppsala, SE-75185, Sweden · Antaros Medical, Mölndal, SE-43153, Sweden

Jun 10, 2026J/K move · Enter open · S save

Time-Conditioned and Multi-Time Survival Prediction from 2D PET/CT Projections in Lung Cancer

Ashish Chauhan, Sambit Tarai, Elin Lundström, Johan Öfverstedt+2

Radiology, Department of Surgical Sciences, Uppsala University, Uppsala, Sweden · National Academic Infrastructure for Supercomputing (NAISS), LinköpingJun University, Linköping, Sweden · Antaros Medical, Mölndal, Sweden+1

Cancer Survival Prediction Survival Analysis PET/CT Imaging Oncology

Open paper AskPDF arXiv

Accurate prediction of overall survival (OS) from positron emission tomography/computed tomography (PET/CT) can support personalized treatment and follow-up strategies in oncology. However, the impact of temporal modeling on imaging-based survival prediction remains insufficiently explored. We investigate how different temporal formulations influence survival prediction by developing two complementary approaches: Attention-guided Time-Conditioned Survival (ATCS) and Multi-Time Survival (MTS). We retrospectively analyzed pre-treatment PET/CT images from 848 patients with non-small cell lung cancer (NSCLC), including 556 for model development and 292 for held-out testing. A previously proposed Time-Conditioned Survival (TCS) model was used as a baseline. Models were trained using 5-fold cross-validation and evaluated on the test set using time-dependent area under the curve (AUC) at 6-month intervals from 0.5 to 5 years. Both ATCS and MTS outperformed the baseline TCS model, achieving mean AUCs of 0.794 and 0.793, respectively, compared to 0.767. ATCS performed better at earlier time points (0.5-3 years), whereas MTS performed better at later intervals (3.5-5 years). Combining tumor-specific and tissue-wise PET/CT features improved performance over either input alone. Finer temporal discretization improved short-term prediction, while coarser intervals provided more stable long-term estimates. These findings demonstrate that temporal modeling and input design influence PET/CT-based survival prediction. The proposed approaches enable time-specific survival estimation from pre-treatment imaging and may support improved risk stratification and clinical decision-making.

Abstract

Mortality Prediction

Explore similar work

Time-Conditioned and Multi-Time Survival Prediction from 2D PET/CT Projections in Lung Cancer

Structured Proxy Features for Multimodal NSCLC Survival Prediction from Pretreatment CT

Time-driven Survival Analysis from FDG-PET/CT in Non-Small Cell Lung Cancer