Multicentre, observational and retrospective study that includes IPF patients diagnosed between 2012 and 2016 that presented FVC greater than or equal to 80% of its theoretical value. The included patients were followed-up every 6 months (clinical and pulmonary functional test evaluation). Patients were recruited from 18 Spanish hospitals that covered the different regions of the whole country, most of them were part of the Spanish IPF Registry or the Observatory IPF.cat. IPF diagnosis was evaluated in a multidisciplinary discussion including expert ILD physicians, pathologists and radiologists on the basis of the respective ATS/ERS/JRS/ALAT guidelines.1 HRCT scan images of all patients were checked by expert ILD radiologists following the same international consensus criteria ATS/ERS and Fleischner Society (only two centers didn’t work with expert ILD radiologist and they sent the anonymized HRCT images to the Multidisciplinary ILD Committee of University Hospital of Bellvitge). All patients provided written informed consent and the study was approved by the Institutional Ethics Committee (CEIC, ref. PR307/16). Data management has followed the regulatory bases included in the EU 2016/679 statement and the Declaration of Helsinki (Fortaleza, 2013). The follow-up was documented for 3 years.

Baseline demographic and clinical characteristics at diagnosis were collected, including age, gender, smoking history, comorbidities, respiratory symptoms (dyspnea, cough) and initiated treatment. Lung function variables (FVC, DLCO, oxygen desaturation in 6MWT) and radiological high-resolution computing tomography (HRCT) pattern at diagnosis were recorded. The HRCT findings and radiological pattern were interpreted on the basis of recommendations from the nomenclature committee of the Fleischner Society.19 Chest HRCT scans were performed on multidetector scanners, volumetric images being acquired with the patient lying in supine position, after maximal inspiration, and without iodinated contrast medium.

The characteristics for each identifiable radiological pattern in IPF were previously defined to improve not only the reading by each radiologist, but also to increase agreement among evaluations (Supplementary Table 1). The presence of honeycombing in those cases with consistent UIP pattern was graded (greater than or equal to 5%). Emphysema was defined as low-attenuation areas bordered by a very thin (<1mm) or no wall in the normal lung. Combined pulmonary fibrosis emphysema syndrome (CPFE) was considered when more than 15% was present.20,21

The annual follow-up data was collected until the end of the study, death, or lung transplantation. FVC and DLCO changes, appearance of new comorbidities such as lung cancer, acute exacerbations (characterized by rapid progression of dyspnea within the previous 30 days, with the presence of new bilateral ground-glass opacities or consolidation by radiography/HRCT without pneumo-thorax or pleural effusion, and a marked decrease in the PaO2),3 lung transplantation and mortality were documented. Disease progression was considered in cases that presented absolute decline of FVC≥10% and/or DLCO≥15% of theoretical value over one year.

Categorical data are described as frequency and percentages, and continuous variables are expressed as mean±standard deviation (SD) or median±interquartile range as appropriate. Group comparisons were made using Chi-squared test or Fisher's exact test when required. Differences in continuous variables were analyzed with ANOVA tests or Student's t test, or their corresponding non-parametrical tests when required (Kruskal–Wallis and Mann–Whitney U tests). Time to event data (time to lung transplant and/or death) were analyzed using Kaplan–Meier analysis. Cox proportional hazard regression for multivariable analyses was used to determine whether the following factors at diagnosis increased the mortality risk: gender, age, smoking history, emphysema, DLCO, exercise capacity, radiological HRCT pattern (model Log-Rank test p-value <0.001; concordance index 0.8). A p-value <0.05 was considered statistically significant. Data were analyzed by using SPSS 24 (SPSS, IBM Corp) and R (software version 3.6.2).

From 225 patients, the majority were male (72.9%), had a mean age of 69.5 years, and 66.5% of them reported exposure to tobacco. The predominant radiological pattern in the HRCT was consistent UIP (51.6%), with an extension of more than 5% of honeycombing in 36.3% of them. Emphysema was present in 33.3% of cases, but only 12.9% fulfilled CPFE criteria. The median FVC was 89.6% (IQR 17) and DLCO was 56.2% (SD 18.5%). Interestingly, 58.7% of patients presented a decrease in DLCO<60% of predicted value and 44.7% of them had associated emphysema.

Only 34 patients (15.1%) were asymptomatic. Of those, 9 (26.5%) presented normal DLCO (>80% predicted value) and indeterminate or probable UIP pattern in HRCT, and only 1 (2.9%) had normal DLCO and consistent UIP pattern. Of the asymptomatic patients, 14 (41.2%) had decreased DLCO (<80% pred.) and indeterminate or probable UIP pattern, and 10 (29.4%) had decreased DLCO and consistent UIP pattern. One subjects from this latter group died during follow-up, but no events were observed in patients with decreased DLCO and indeterminate or probable UIP.

Regarding IPF treatment, 124 (55.1%) of patients received anti-fibrotic medication at diagnosis, 46 cases nintedanib and 59 pirfenidone (in 19 patients the type of anti-fibrotic drug was missing).

53.3% of patients presented disease progression at the 3-years of follow-up; of them 14.7% died or received lung-transplant within 3 years after diagnosis. Furthermore, from this group only 7 patients underwent lung transplant, the other 26 cases died.

The cause of death was due to IPF disease progression in most cases (74.8%), lung cancer in 1% and to other reasons not related to the disease in 24.2%. Only 2% of patients presented an acute exacerbation that required hospitalization. When comparing the group who died within 3 years with those who were still alive, we found differences between the two groups (Table 2).

The patients who died were more symptomatic at diagnosis, and the majority presented dyspnea (93.9%) and cough (71.8%). All patients with no symptoms, normal DLCO and indeterminate or probable UIP pattern at diagnosis were alive at 3 years.

It should be noted that the group of patients that died had a lower DLCO at diagnosis (mean 42.1%±16 vs. 58.7%±17.9; p<0.001). When analyzing the basal DLCO in three degrees of severity (mild: DLCO 80–60%, moderate: 60–40%, severe: <40%22), we observed that patients presenting severe DLCO decrease had a worse prognosis and higher risk of mortality and/or lung transplantation (Fig. 1). Up to 81.8% of patients that died presented moderate-severe decrease of DLCO. Patients with severe DLCO reduction at diagnosis presented disease progression in 61.4% cases and a 3-year mortality of 20.4%.

Fig. 1.

Survival by DLCO.

(0.14MB).

Oxygen desaturation below 90% during the 6MWT was also more frequently present at diagnosis among those cases that died (70%) than in the group that was alive (30.5%).

Regarding the chest HRCT pattern at diagnosis, we found remarkable differences. In the deceased group, the predominant pattern had been consistent UIP (91.7%). On the other hand, among those alive, the radiological patterns were more variable; 54.3% had a pattern of UIP, 41% probable UIP, and the pattern of indeterminate UIP was similar in both groups (4%). Among those cases with consistent UIP, the extension of honeycombing >5% was very frequent in the group of deaths (72.7%) when comparing with those that were alive (29.8%), p<0.001 (Fig. 2).

Fig. 2.

Survival by HRCT.

(0.13MB).

A higher proportion of former or current smokers was present in the group that died or underwent to lung transplant compared with the alive group after 3-years (86.9% vs. 63.5%, p=0.046). The presence of emphysema was not statistically different (0.075) and neither the CPFE cases. However, 15 of 16 IPF patients with emphysema that died met CPFE criteria and the mortality rate among CPFE cases was three times greater (51.7%) than in the global cohort (14.7%), which differs from those emphysema cases without CPFE criteria (2.17%). Therefore, the poor prognosis associated with smoking history could be attributed in part to those cases in which this habit induces CPFE.

Regarding anti-fibrotic medication, 48.5% of patients who were deceased at 3 years had received anti-fibrotic treatment from diagnosis, while 56.2% of patients that were alive had been treated from the beginning.

A Cox proportional hazard model was constructed with objective variables of clinical relevance and statistically significant on the univariate analyses, including age, gender, smoking habit, oxygen desaturation in the 6MWT, DLCO (categorized as mentioned above) and presence of UIP radiological pattern. The model showed that patients with a very severe DLCO decrease and consistent UIP radiological pattern had a higher risk for mortality (HR 7.4, 95% CI 1.06–51.9, p=0.043; and HR 5.2, 95% CI 1.13–23.5, p=0.034, respectively). Results for this model are shown in Fig. 3.

Fig. 3.

A Cox proportional model.

(0.17MB).

According to these results, the prognostic factors at diagnosis of IPF with preserved FVC were the presence of consistent UIP pattern in chest HRCT, the severe decrease in DLCO.

Considering IPF management depending on disease behavior, patients with preserved FVC at higher risk for 3-year mortality or lung transplant would not be considered early or mild IPF (including the presence of radiological UIP consistent pattern and CPFE criteria). Regarding the identified factors associated with disease behavior in these patients, we differentiated stage 0 or “early IPF” from the other IPF patient profiles with preserved FVC (Table 3). Stage 0 would include a minority of cases without symptoms that present with normal DLCO and indeterminate or probable UIP pattern in the chest HRCT. In our cohort, only 9 (4%) met these criteria. All of them were GAP stage I. Only one stage 0 patient showed disease progression, but none had died or received lung-transplant at 3 years of follow-up. Median FVC for stage 0 patients was 105.4% (IQR 18), and mean DLCO was 93% (SD 13.6%). Regarding chest HRCT pattern, six patients showed probable UIP, and 3 indeterminate UIP. Lung biopsy was required in 44.4% of these cases; 3 of them with morphological changes of consistent UIP and 1 probable UIP.

On the other hand, non-early IPF patients would be characterized by normal FVC (≥80%) but decreased DLCO (<80%) and/or consistent or probable UIP and/or presence of respiratory symptoms (Table 3). Contrary to what was observed for stage 0, 119 (54.3%) subjects showed disease progression, and 33 (15%) died or required lung transplant at 3 years.