Pulmonary Symptoms Measured by the NIH Lung Score Predict Overall Survival and Non-Relapse Mortality in Chronic Gvhd

The 2005 NIH Consensus Conference recommended assessment of lung function in patients with chronic graft-versus-host disease (cGVHD) by both pulmonary function tests (PFTs) and assessment of pulmonary symptoms. The Chronic GVHD Consortium has made serial collections of these measures. Based on prior literature, we hypothesized that the following subgroups would have higher non-relapse mortality (NRM) or lower overall survival (OS): 1. Obstructive lung disease; 2. Higher NIH lung score based on symptoms; 3. Higher Lung Function Score (LFS) based on PFTs; 4. Clinical diagnosis of bronchiolitis obliterans syndrome (BOS). Methods:  Patients ≥ 2 years of age diagnosed with cGVHD requiring systemic treatment ≤ 3 years after transplantation were assessed every 3-6 months. Available PFT data were abstracted, including FEV1, forced vital capacity (FVC) and diffusing capacity of lung for carbon monoxide (DLCO). Obstructive disease was defined as a decreased FEV1 (tested at <50%, <75%, <80%) and FEV1/FVC < 0.70. The NIH lung score of 0-3 was based on the severity of pulmonary symptoms with exertion. The LFS was based on FEV1 and DLCO. A clinical diagnosis of BOS was collected from the clinician-completed data capture form. Cox regression models were fit for OS and NRM using a time-varying covariate model for lung function measures and adjusting for patient and transplant characteristics and non-lung cGVHD severity. Results: There were 1591 visits used in this analysis, only 845 (50%) of the visits had recorded PFTs. Obstructive physiology (FEV1<80%) was found in 184 visits and 54 visits (6%) had FEV1 <50%.  BOS was reported on 122 visits.   Median follow up time was 20.1 mo (range 2.9 – 47.7), and median survival has not been reached. NIH lung score based on symptoms was strongly correlated with NRM and OS (Table). The NIH lung score correlated with Karnofsky performance status (correlation = -0.34), and both had independent prognostic significance in multivariate models. None of the other measures including obstructive PFTs, LFS, or the clinician's indicator of BOS, were significantly associated with OS or NRM.  Conclusion: NIH clinical lung score of 0-3 based on pulmonary symptoms is predictive of OS and NRM in patients with cGVHD.  PFT derived data, including the LFS and obstructive physiology, and a clinical diagnosis of BOS are not associated with OS or NRM. Our conclusions are limited by inconsistent practices in performing PFTs.

Table: Relationship between NIH Lung score based on symptoms and OS and NRM

	Overall survival					Non-relapse mortality
	overall p	Value	n	HR, 95% CI	p	overall p	Value	n	HR, 95% CI	p
NIH Clinical Lung Score based on Symptoms (0-3)	0.007	0	1138	1.0		0.01	0	1132	1.0
		1	299	2.24 (1.35-3.72)	0.002		1	292	2.34 (1.24-4.43)	0.009
		2	94	1.94 (0.90-4.19)	0.09		2	93	2.55 (1.03-6.30)	0.04
		3	22	3.70 (1.04-13.17)	0.04		3	22	5.17 (1.34-19.87)	0.02