京都大学 / 0048 / 新制・課程博士 / 博士(医学) / 甲第21954号 / 医博第4496号 / 新制||医||1037(附属図書館) / 京都大学大学院医学研究科医学専攻 / (主査)教授 渡邉 大, 教授 伊佐 正, 教授 椛島 健治 / 学位規則第4条第1項該当 / Doctor of Medical Science / Kyoto University / DFAM

京都大学 / 0048 / 新制・課程博士 / 博士(医学) / 甲第22320号 / 医博第4561号 / 新制||医||1041(附属図書館) / 京都大学大学院医学研究科医学専攻 / (主査)教授 椛島 健治, 教授 大森 孝一, 教授 森田 智視 / 学位規則第4条第1項該当 / Doctor of Medical Science / Kyoto University / DFAM

Background Idiopathic pulmonary fibrosis (IPF) is a devastating form of chronic lung injury of unknown aetiology characterised by progressive lung scarring. A diagnosis of definite IPF requires High Resolution Computed Tomography (HRCT) appearances indicative of usual interstitial pneumonia (UIP), or in patients with 'possible UIP' CT appearances, histological confirmation of UIP. However the proportion of such patients that undergo SLB varies, perhaps due to a perception of risk of biopsy and additive diagnostic value of biopsy in individual patients. We hypothesised that an underlying UIP pathological pattern may result in increased risk of death and aimed to explore this by comparing the risk of SLB in suspected idiopathic interstitial pneumonia, stratified according to HRCT appearance. Additionally we sought to determine the positive-predictive value of biopsy to diagnose IPF in patients with 'possible UIP HRCT' in our population. In patients with possible UIP who are not biopsied, the clinical value of bronchoalveolar lavage (BAL) is uncertain. We aimed to prospectively study the diagnostic and prognostic value of BAL differential cell count (DCC) in suspected IPF and determine the feasibility of repeat BAL and the relationship between DCC and disease progression in two successive BALs. We hypothesised that BAL DCC between definite and possible IPF was different and that baseline DCC and change in BAL DCC predicted disease progression. Alveolar macrophages (AMs) are an integral part of the lung's reparative mechanism following injury, however in IPF they contribute to pathogenesis by releasing pro-fibrotic mediators promoting fibroblast proliferation and collagen deposition. Expansion of novel subpopulations of pulmonary monocyte-like cells (PMLCs) has been reported in inflammatory lung disease. We hypothesised that a distinct AM polarisation phenotype would be associated with disease progression. We aimed to perform detailed phenotyping of AM and PMLCs in BAL in IPF patients. Several prognostic scoring systems and biomarkers have been described to predict disease progression in IPF but most were derived from clinical trial patients or tertiary referral centres and none have been validated in separate cohorts. We aimed to identify a predictive tool for disease progression utilising physiological, HRCT and serum biomarkers in a unique population of incident treatment naïve IPF patients. Methods Between 01/01/07 and 31/12/13, 611 consecutive incident patients with suspected idiopathic interstitial pneumonia (IIP) presented to the Edinburgh lung fibrosis clinic. Of these patients 222 underwent video-assisted thoracoscopic lung biopsy and histological pattern was determined according to ATS/ERS criteria. Post-operative mortality and complication rates were examined. Fewer than 2% received IPF-directed therapy and less than 1% of the cohort were lost to follow-up. Disease progression was defined as death or ≥10% decline in VC within 12 months of BAL. Cells were obtained by BAL and a panel of monoclonal antibodies; CD14, CD16, CD206, CD71, CD163, CD3, CD4, CD8 and HLA-DR were used to quantify and selectively characterise AMs, resident PMLCs, inducible PMLCs, neutrophils and CD4+/CD8+ T-cells using flow cytometry. Classical, intermediate and non-classical monocyte subsets were also quantified in peripheral blood. Potential biomarkers (n=16) were pre-selected from either previously published studies of IPF biomarkers or our hypothesis-driven profiling. Linear logistic regression was used on each predictor separately to assess its importance in terms of p-value of the associated weight, and the top two variables were used to learn a decision tree. Results Based on the 2011 ATS/ERS criteria, 87 patients were categorised as 'definite UIP', of whom 3 underwent SLB for clinical indications. IPF was confirmed in all 3 patients based on 2013 ATS/ERS/JRS/ALAT diagnostic criteria. 222 patients were diagnosed with 'possible UIP'; 55 underwent SLB, IPF was subsequently diagnosed in 37 patients, 4 were diagnosed with 'probable IPF' and 14 were considered 'not IPF'. In this group, 30 patients were aged 65 years or over and 25/30 (83%) had UIP on biopsy. 306 patients had HRCTs deemed 'inconsistent with UIP', SLB was performed in 168 patients. Post6 operative 30-day mortality was 2.2% overall, and 7.3% in the 'possible UIP' HRCT group. Patients with 'definite IPF' based on HRCT and SLB appearances had significantly better outcomes than patients with 'definite UIP' on HRCT alone (P=0.008, HR 0.44 (95% CI 0.240 to 0.812)). BAL DCC was not different between definite and possible UIP groups, but there were significant differences with the inconsistent with UIP group. In the 12 months following BAL, 33.3% (n=7/21) of patients in the definite UIP group and 29.5% (n=18/61) in the possible UIP group had progressed. There were no significant differences in BAL DCC between progressor and non-progressor groups. Mortality in patients with suspected IPF and a BAL DCC consistent with IPF was no different to those with a DCC inconsistent with IPF (P=0.425, HR 1.590 (95% CI 0.502 to 4.967)). There was no difference in disease progression in either group (P=0.885, HR 1.081 (95% CI 0.376 to 3.106)). There was no statistically significant difference in BAL DCC at 0 and 12 months in either group. There was no significant change in DCC between 0 and 12 month BALs between progressors and non-progressors. Repeat BAL was well tolerated in almost all patients. There was 1 death within 1 month of a first BAL and 1 death within 1 month of a second BAL; both were considered 'probably procedure-related'. AM CD163 and CD71 (transferrin receptor) expression were significantly different between groups (P < 0.0001), with significant increases in the IPF group vs non fibrotic ILD (P < 0.0001) and controls (P < 0.0001 and P < 0.001 respectively). CD71 expression was also significantly increased in the IPF progressor vs non-progressor group (P < 0.0001) and patients with high CD71 expression had significantly poorer survival than the CD71low group (P=0.040, median survival 40.5 and 75.6 months respectively). CD206 (mannose receptor) expression was also significantly higher in the IPF progressor vs non-progressor group (P=0.034). There were no differences in baseline BAL neutrophil, eosinophil or lymphocyte percentages between IPF progressor or non-progressor groups. The percentage of rPMLCs was significantly increased in BAL fluid cells of IPF patients compared to those with non-fibrotic ILD (P < 0.0001) and healthy controls (P < 0.05). Baseline rPMLC percentage was significantly higher in IPF progressors vs IPF non-progressors (P=0.011). Baseline BAL iPMLC:rPMLC ratio was also significantly different between IPF progressor and non-progressor groups (P=0.011). Disease progression was confidently predicted by a combination of clinical and serological variables. In our cohort we identified a predictive tool based on two key parameters, one a measure of lung function and one a single serum biomarker. Both parameters were entered into a decision tree, and when applied to our cohort yielded a sensitivity of 86.4%, specificity of 92.3%, positive predictive value of 90.5% and negative predictive value of 88.9%. We also applied previously reported predictive tools such as the GAP Index, du Bois score and CPI Index to the Edinburgh IPF cohort. Conclusions SLB can be of value in the diagnosis of ILD, however perhaps due to the perceived risks associated with the procedure, only a small percentage of patients undergo SLB despite recommendations that patients have histological confirmation of the diagnosis. Advanced age is a strong predictor for IPF, and in our cohort 83% of patients aged over 65 years with 'possible UIP' HRCT appearances, had UIP on biopsy. BAL and repeat BAL in IPF is feasible and safe (< 1.5% mortality). Of those that underwent repeat BAL, disease progression was not associated with a change in DCC. However, 22% of lavaged patients died or were deemed too frail to undergo a second procedure at 12 months. These data emphasise the importance of BAL in identifying a novel human AM polarisation phenotype in IPF. Our data suggests there is a distinct relationship between AM subtypes, cell-surface expression markers, PMLC subpopulations and disease progression in IPF. This may be utilised to investigate new targets for future therapeutic strategies. / Disease progression in IPF can be predicted by a combination of clinical variables and serum biomarker profiling. We have identified a unique prediction model, when applied to our locally referred, incident, treatment naïve cohort can confidently predict disease progression in IPF. IPF is a heterogeneous disease and there is a definite clinical need to identify 'personalised' prognostic biomarkers which may in turn lead to novel targets and the advent of personalised medicines.