Surfactant protein-D predicts prognosis of interstitial lung disease induced by anticancer agents in advanced lung cancer: a case control study

Between April 2011 and March 2016, 1437 patients were diagnosed with advanced lung cancer at Juntendo University Hospital and Juntendo University Urayasu Hospital, and 1224 of them received chemotherapy, including cytotoxic agents and epidermal growth factor receptor (EGFR) tyrosine kinase inhibitors. Thirty-six of these patients diagnosed with ILD-AA (13 at Juntendo University Hospital and 23 at Juntendo University Urayasu Hospital) were enrolled in this case control study. All patients were diagnosed with ILD-AA on the basis of HRCT findings and elevated levels of serum markers, including KL-6 and SP-D. At the time of diagnosis with ILD-AA, there was no evidence of infection, heart failure, or renal failure. Patients receiving immunotherapy, operation, and thoracic radiotherapy were excluded from the study. Patients who died within 6 weeks of the onset of ILD-AA were included in the death group and those who survived over 6 weeks were included in the survival group (Fig. 1). The study protocol was approved by the Juntendo University Ethical Committee and registered under number 16-051. Owing to the retrospective nature of the research, the Ethical Committee waived the requirement for informed consent.

The serum biomarkers KL-6 and SP-D were measured in all of the patients prior to the diagnosis of ILD-AA and at the time of ILD-AA diagnosis. Serum KL-6 and SP-D were measured using a sandwich enzyme-linked immunosorbent assay, with KL-6 and SP-D antibodies (SRL, Inc. Tokyo, Japan). The cut off values for serum KL-6 and SP-D were set at 500 U/mL and 110 ng/mL, respectively. The differences (ΔKL-6 and ΔSP-D) were calculated as the value of KL-6 and SP-D levels at the onset of ILD-AA minus the respective values prior to ILD-AA.

At the onset of ILD-AA, all of the patients underwent HRCT, which was performed with 2-mm collimation at 10-mm intervals, from the lung apex to the lung base. Based on the HRCT results at the time of diagnosis with ILD-AA, patients were classified into four groups: HP, DAD, CIP, and organized pneumonia/eosinophilic pneumonia (OP/EP). HP was defined as only ground glass opacity (GGO) without traction bronchiectasis and honeycombing. DAD was characterized by extensive bilateral GGO or consolidation with traction bronchiectasis. CIP was defined as evidence of fibrosis, including subpleural reticular shadow, honeycombing, or reticular shadows on the bronchovascular bundles. OP/EP was characterized by peribronchial and/or subpleural consolidation of the bronchovascular bundles.

We evaluated KL-6 and SP-D at the onset of ILD-AA, ΔKL-6 and ΔSP-D, the risk factors for survival and death, the HRCT findings, and the effect of high or low ΔKL-6 and ΔSP-D on survival.

Thirty-six patients diagnosed with ILD-AA during treatment with anticancer agents were enrolled in this study. Patient’s characteristics are shown in Table 1. All of the patients were Japanese. Patient’s median age was 71 years (range: 53-87 years). Nine (25%) patients were women, 31 (86.1%) were smokers, 27 (75%) had good performance status (PS = 0, 1), 21 (58.3%) had pre-existing interstitial shadow on HRCT, and 23 (63.9%) had emphysema on HRCT. Twenty-four patients had adenocarcinoma, nine had small cell carcinoma, and three had squamous cell carcinoma. Five patients had sensitive EGFR mutation. No patients had the EML4/ALK fusion gene. All 36 patients received several types of chemotherapy regimens. Suspected regimens are shown as Table 2. Patients without the EGFR mutation were treated with cytotoxic chemotherapy, including pemetrexed plus platinum agents ( n = 7), pemetrexed monotherapy ( n = 3), carboplatin plus paclitaxel ( n = 4), albumin combined with paclitaxel ( n = 1), docetaxel ( n = 5), bevacizumab with carboplatin plus paclitaxel ( n = 2), etoposide plus platinum agents ( n = 4), amurubicin ( n = 3), and nogitecan ( n = 2). There was no association between death related to ILD-AA and any specific anticancer agent. Moreover, there was no evidence of cancer progression or carcinomatous lymphangitis by HRCT findings and tumor marker elevation in all patients. Fourteen patients died of respiratory failure related to ILD-AA within 6 weeks.

Serum KL-6 and SP-D levels were analyzed prior to and at the onset of ILD-AA in all 36 patients. The median serum KL-6 and SP-D values prior to ILD-AA were 1100 U/mL (range: 327–3328 U/mL) and 314 ng/mL (range: 22–393 ng/mL), respectively. When compared to the values prior to ILD-AA, serum KL-6 levels in 31 patients (86.1%) and serum SP-D in all patients (100%) at the onset of ILD-AA were increased from prior to ILD-AA. Figure 2 shows the serum KL-6 and SP-D levels at the onset of ILD-AA by outcome. When patients in the survival group were compared to those in the death group, the serum SP-D levels in the death group at the onset of ILD-AA were significantly higher than those in the survival group (Mann–Whitney U-test; p = 0.002, Fig. 2b). However, serum KL-6 levels did not differ significantly between the two groups ( p = 0.833, Fig. 2a). When ΔSP-D was compared between the two groups, ΔSP-D in the death group was significantly higher than that in the survival group ( p = 0.0008, Fig. 2d); however, ΔKL-6 did not differ considerably between the two groups ( p = 0.282, Fig. 2c).

We also evaluated the association between HRCT patterns at the onset of ILD-AA and serum biomarkers. In the study population, HRCT patterns were as follows: HP ( n = 6), DAD ( n = 14), CIP ( n = 7), and OP/EP ( n = 9). Figure 3a and b show ΔKL-6 and ΔSP-D according to ILD-AA patterns. As shown in Fig. 3a, ΔKL-6 in patients with the DAD pattern was significantly higher than in patients with the HP or OP/EP patterns (DAD-HP , p = 0.019; DAD-OP/EP , p = 0.033); however, there was no significant difference in ΔKL-6 between the DAD and CIP ( p = 0.962). In patients with DAD, ΔSP-D was significantly elevated when compared to the three other types of ILD-AA (DAD-HP, p = 0.011; DAD-CIP, p = 0.022; DAD-OP/EP, p = 0.029; Fig. 3b); however, there were no significant differences in ΔSP-D among the three other types of ILD-AA. Therefore, ΔSP-D was significantly related to the DAD pattern. Although 21 patients had pre-existing interstitial shadow on HRCT, there was no significant difference between HRCT patterns and pre-existing interstitial shadow. Moreover, pre-existing interstitial shadow in all patients were categorized into IIPs. Of these patients, 14 patients with pre-existing interstitial shadow were clinically diagnosed with idiopathic pulmonary fibrosis (IPF). The subtypes of IIPs in pre-existing interstitial shadow were not associated with ILD-AA patterns. Then, of the 14 patients with IPF, the serum SP-D levels in 5 patients and serum KL-6 levels in 12 patients were elevated before the diagnosis of ILD-AA. However, there was no significant association between change in serum markers and pre-existing IPF (data not shown). Moreover, there were no findings of carcinomatous lymphangitis, including the thickening of the bronchovascular bundle, interlobular septa, and centrilobular micro nodules on HRCT in any patients.

Serum KL-6 and SP-D levels were also measured 2 weeks after the onset of ILD-AA in 12 patients in the survival group (50%) and seven patients in the dead group (58.3%). Changes in KL-6 and SP-D values between these two time points (2 weeks after the onset of ILD-AA and the onset of ILD-AA) are shown in Fig. 4. Changes in SP-D levels in the 12 patients in the survival group were significantly decreased when compare to those from the seven patients in the dead group (−126 ± 67.91 ng/mL in the survival group versus 284 ± 114.30 ng/mL in the dead group, p = 0.033). In contrast, the change in KL-6 was not significantly different between two groups (137 ± 147.62 U/mL in the survival group versus 314 ± 208.77 U/mL in the dead group, p = 0.752).

To obtain optimal cut off values for ΔKL-6 and ΔSP-D in serum for prognostic assessments in patients with ILD-AA, a ROC curve analysis was performed using the highest ΔKL-6 and ΔSP-D values (Fig. 5a). To predict the risk of mortality within 6 weeks of the onset of ILD-AA, the optimal cut off value for ΔSP-D was 398 ng/mL. The cut off value of ΔKL-6 was nonsignificant because of a low likelihood ratio and area under the curve. Eight of the nine (88.9%) patients with high ΔSP-D (≥ 398 ng/mL) died, and six of the 27 (22.2%) patients with low ΔSP-D (< 398 ng/mL) died ( p = 0.0003).

Median survival time (MST) was 93 days in all of the patients diagnosed with ILD-AA (95% CI, 36–174; Fig. 5b). Survival time for the patients with low ΔSP-D was significantly longer than that for the patients with high ΔSP-D (MST, 159 days; 95% CI, 72–328 in low ΔSP-D versus MST, 30 days, 95% CI, 3–33 in high ΔSP-D, HR: 26.02, p = 0.001, Fig. 5c).

The results of univariate and multivariate analyses of the risk factors for death associated with ILD-AA are shown in Tables 3 and 4. In univariate analysis, high ΔSP-D and smoking history were significantly associated with ILD-AA-related death (high ΔSP-D: odds ratio [OR], 7.00; 95% CI, 2.19–72.26; p = 0.001 and smoking history: OR, 2.48; 95%CI, 1.26–4.86; p = 0.042, respectively). Multivariate analysis performed using six variables (age, smoking history, performance status, the presence of emphysema, the presence of interstitial shadow, and high ΔSP-D) showed that only high ΔSP-D was a significant independent risk factor for ILD-AA-related death (OR, 25.56; 95% CI, 2.29–285.46; p = 0.008).