Increased monocyte level is a risk factor for radiological progression in patients with early fibrotic interstitial lung abnormality

Background Interstitial lung abnormalities (ILA) are specific spatial patterns on computed tomography (CT) scan potentially compatible with early interstitial lung disease. A proportion will progress; management involves risk stratification and surveillance. Elevated blood monocyte levels have been shown to associate with progression of idiopathic pulmonary fibrosis. The aims of the present study were: 1) to estimate the proportion of “early fibrotic” (EF)-ILAs (reticular±ground-glass opacities, excluding traction bronchiectasis and honeycombing) on CT scans of patients attending all-indications thoracic CTs, and proportion demonstrating radiological progression; and 2) to explore association between peripheral blood leukocyte levels and ILA progression. Methods We analysed all thoracic CT reports in individuals aged 45–75 years performed between January 2015 and December 2020 in one large teaching hospital (Oxford, UK) to identify patient CT reports consistent with EF-ILA. CT-contemporaneous blood leukocyte counts were examined to explore contribution to progression and all-cause mortality, using multivariate Cox regression. Results 40 711 patients underwent thoracic CT imaging during this period. 1259 (3.1%) demonstrated the EF-ILA pattern (mean±sd age 65.4±7.32 years; 735 (47.8%) male). EF-ILA was significantly associated with all-cause mortality (hazard ratio 1.87, 95% CI 1.25–2.78; p=0.002). 362 cases underwent at least one follow-on CT. Radiological progression was observed in 157 (43.4%) cases: increase in reticulation n=51, new traction bronchiectasis n=84, honeycombing n=22. Monocyte count, neutrophil count, monocyte:lymphocyte ratio, neutrophil:lymphocyte ratio and “systemic inflammatory response index” were significantly associated with radiological progression. Conclusion 3.1% of subjects requiring thoracic CT during a 6-year period demonstrated EF-ILA. Monocyte levels and blood leukocyte-derived indexes were associated with radiological progression and could indicate which patients may require closer follow-up.

Search criteria were defined with close guidance from our lead thoracic radiologist (RB) based upon the vocabulary tendencies of our thoracic radiologists when reporting CTs. We defined upper limit of age as 75 years. 1 Reticular abnormalities are common in older individuals and have sometimes been regarded as part of the normal spectrum of senescent lung. 2 Reporting findings as such could downgrade its clinical significance and introduce ambiguity to this dataset. 3 Possible UIP was included in search criteria as this term, used in the 2011 iteration of the IPF guideline, can also include reticulation in absence of traction bronchiectasis and distinct absence of honeycombing. 4 Fibrotic ILAs with basal and peripheral predominance are considered to possess greater risk of progression and mortality. As such this distribution was also incorporated into the search criteria in an "OR" rather than "AND" fashion. This was to maximise the initial search results and also because it has previously been documented that under-reporting of ILA has previously been described. 5 In Phase 2, qualitative information was extracted from CT reports (identified from above search criteria) and converted into structured binary data. Additional parenchymal features were screened for in CT scan reports, adjudicated and collated. Features included ground glass opacities (GGO), emphysema, traction bronchiectasis and honeycombing. 6,7 To ensure we . Specific combinations of concurrent parenchymal features (e.g. reticulation and GGO) were also collated. Importantly, traction bronchiectasis and honeycombing are considered representative of established fibrotic ILD and therefore cases with co-existing traction bronchiectasis and honeycombing identified in phase 2 were partitioned from cases of EF-ILA. Where feasible cases with reticulation and or traction bronchiectasis and or honeycombing were sub-classified into UIP categories to communicate ILA extent and co-existence to the readership. 8 This does not assume cases categorised in this manner have IPF. Similarly, cases with reticulation were subcategorised into cases with and without other non-fibrotic parenchymal abnormalities including GGO. Cases identified from the preliminary key word search (phase 1), however later found on screening (phase 2) that CT reports were detailing negative / absence of specific radiological patterns were defined as a ' Nil ILA' reference cohort.
Radiographic progression was recorded as a binary event based on presence of new or increase in pre-existing parenchymal features. It was captured based upon text-based searching of CT report descriptions. Progression was adjudicated based upon perusal of all CT reports of repeat CTs by one specialist trainee (AA) using pre-defined criteria. Progression was defined as increase in either extent of identified early fibrotic features (reticulation and or GGO), new emergence of traction bronchiectasis, and/or new emergence of honeycombing on follow on CT. In cases not demonstrating radiographic progression, this was defined as unchanged pre-existing parenchymal features and absence of new features.
Total number of subjects that underwent a thorax-protocolised CT between January-2015 and December-2020 was used as a denominator to estimate EF-ILA prevalence in cases indicated to undergo thoracic CT.

CT reporting
CT scans were assessed using text-based searching of CT reports. 80.9% of CTs were reported by 1 of 8 thoracic radiologists. All were consultants (with UK Certificate of Completion of Training/CCT for Radiology) and 7 of 8 of these consultants also held a post -CCT Thoracic Radiology Fellowship and participated in dedicated Interstitial Lung Disease MDTs. The other thoracic radiologist has >20 years of experience in reporting HRCT for ILD in our centre. The remainder were reported by one of 14 Oxford-based post CCT radiologists. Our thoracic radiologists have contributed to >20 radiologybased studies over the last decade. In one study, performed with our research group, the agreement between of reporting ILD features between the two radiologists was excellent (r=0.91; p<0.001 by Pearson's Correlation, also tested by Bland Altman; Suppl Fig 4A). 9

Comorbidity profiles
Comorbidities for each case were search for using ICD-10 coding. Specific ICD-10 codes, 10 pertaining to respiratory and non-respiratory diagnoses were cross-referenced with electronic health records. Similarly, attendance at ILD clinic, date thereof and date of mortality obtained and time intervals between these events and CT date deduced. Comorbidities are listed as binary events. It was not possible to obtain date of comorbidity diagnoses for all cases and therefore not possible to determine which diagnoses were established prior to, at or after first CT scan showing ILA. For this reason, comorbidities were not included in all multivariate analysis and interpretation of this data is limited to association rather than causality.

Statistical analysis
Cox proportional hazard models were used to determine hazard ratios (HRs) to progression and allcause mortality (separate models). In both models, age, gender, monocytes, neutrophils, and lymphocytes levels obtained at a time point closest to the CT scan were included. CoV values for each case were calculated from available counts for monocyte, neutrophil, and lymphocytes (and derived indexes) in the 1 year up to first CT to account for within-group variance in these measures. In separate models, CoV of longitudinal measurements for each leukocyte / index (for each subject) was included in regression models, along with the leukocyte value closest to CT. This was to adjust for any effect that variation in longitudinal measurement of these leukocyte parameters may have on clinical outcome, relative to the corresponding leukocyte value closest to CT used in the model.

Table S2
Multivariate cox regression examining association between blood leukocyte and (i) mortality in "early fibrotic" ILA (n=1259) and (ii) radiological progression in "early fibrotic" ILA cohort with available repeat CT for comparison (n=362). Leukocytes adjusted for age, gender and respiratory comorbidities of COPD, lung cancer and pneumonia. Multivariate cox regression examining association between blood leukocyte indexes and (i) mortality in "early fibrotic" ILA (n=1259) and (ii) radiological progression in "early fibrotic" ILA cohort with available repeat CT for comparison (n=362). Leukocytes adjusted for age, gender and respiratory comorbidities of COPD, lung cancer and pneumonia. Multivariate cox regression examining association between blood leukocyte indexes and (i) mortality in "early fibrotic" ILA (n=1259) and (ii) radiological progression in "early fibrotic" ILA cohort with available repeat CT for comparison (n=362). Covariates in "full blood count", MLR, NLR and SIRI models adjusted for age, gender, and leukocyte co-efficient of variation (CoV).