Transcription

Furukawa et al. Respiratory Research (2017) 18:18DOI 10.1186/s12931-017-0503-3RESEARCHOpen AccessThe St. George’s Respiratory Questionnaireas a prognostic factor in IPFTaiki Furukawa1, Hiroyuki Taniguchi1*, Masahiko Ando2, Yasuhiro Kondoh1, Kensuke Kataoka1, Osamu Nishiyama3,Takeshi Johkoh4, Junya Fukuoka5, Koji Sakamoto6 and Yoshinori Hasegawa6AbstractBackground: It is unclear whether health related quality of life (HRQL) may have a predictive value for mortality inidiopathic pulmonary fibrosis (IPF).We investigated the relationship between HRQL assessed using the St. George’s Respiratory Questionnaire (SGRQ)and survival time in patients with IPF, and tried to determine a clinical meaningful cut off value to predict poorersurvival rates.Methods: We retrospectively analyzed consecutive patients with IPF who underwent an initial evaluation from May2007 to December 2012. The diagnosis of IPF was made according to the 2011 international consensus guidelines.We used Cox proportional hazard models to identify independent predictors for mortality rate in patients with IPF.Results: We examined 182 eligible cases, average age was 66 years old, and 86% were male. Mean levels of percentpredicted FVC, DLco, six-minute-walk test distance, and the SGRQ total score were around 80%, 58%, 580 m, and 34points. On multivariate analysis, the SGRQ total score (hazard ratio [HR], 1.012; 95% confidence interval [CI] 1.001–1.023;P .029) and percent predicted FVC (HR, 0.957; 95% CI 0.944–0.971, P .001) were independent predictors for mortalityrate. Moreover, a score higher than 30 points in the SGRQ total score showed higher mortality rate (HR, 2.047; 95% CI,1.329–3.153; P .001).Conclusions: The SGRQ total score was one of independent prognostic factors in patients with IPF. Total scoreshigher than 30 points were associated with higher mortality rates.Trial registration: This study was retrospective, observational study, so it is not applicable.Keywords: Health related QoL, Idiopathic pulmonary fibrosis, Prognostic factors, The St. George’s RespiratoryQuestionnaireBackgroundIdiopathic pulmonary fibrosis (IPF) is a fatal lung diseasecharacterized by chronic, progressive fibrosing interstitial pneumonia of unknown etiology [1]. As the diseasecondition progresses, dyspnoea on exertion becomessevere and health-related quality of life (HRQL) seriouslydeteriorates [2].HRQL is a subjective and multidimensional appraisalthat focuses on the impact of illness and treatment onphysical, emotional, and social well-being [3]. HRQL toolscan thus capture various information that physiologic or* Correspondence: [email protected] of Respiratory Medicine and Allergy, Tosei General Hospital, 160Nishioiwake-cho, Seto, Aichi 489-8642, JapanFull list of author information is available at the end of the articleradiologic measures cannot. Therefore, it has been considered to be one of the most important endpoints inpharmacological trials and rehabilitation for IPF [4, 5].The St. George’s Respiratory Questionnaire (SGRQ) isthe most frequently used tool to measure HRQL. It wasdeveloped as a standardized, self-administered, diseasespecific health status evaluation scale for patients withchronic obstructive disease [6]. The questionnaire consists of 50 items with 76 weighted responses that produce scores in three domains and one total score. Thedomains are symptoms (breathlessness, cough, andwheeze), activities (that are limited by the symptoms),and impacts (social and psychologic effect of the respiratory diseases). The score for each domain is calculated The Author(s). 2017 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, andreproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link tothe Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication o/1.0/) applies to the data made available in this article, unless otherwise stated.

Furukawa et al. Respiratory Research (2017) 18:18from 0 to 100, with higher scores corresponding toworse HRQL.In patients with COPD, the SGRQ was shown to be asignificant predictor of mortality [7, 8]. However, inpatients with IPF, the relationship between HRQLassessed using the SGRQ and mortality has not fullybeen studied.The aim of the present study was to investigatewhether the SGRQ has a predictive value for mortalityin IPF, including various dimensional clinical factors previously reported as prognostic factors, namely MMRC[9], % predicted FVC, % predicted DLco [1, 10–15],6MWD, desaturation during 6MWT [13, 16–18], andBMI [19]. We also sought a clinically meaningful cut offvalue to predict poorer mortality rate.MethodsStudy subjectsTwo-hundred and five consecutive patients with IPF,who underwent systemic initial evaluation from May2007 to December 2012 at a center for respiratory diseases (Aichi, Japan), were identified from their medicalrecords. The diagnosis of IPF was made according to the2011 international consensus guidelines, and since 2011the accuracy of the diagnosis has been confirmed bymultidisciplinary discussion (MDD); however, some ofthe patients who initially presented before 2010 hadbeen diagnosed based on different guidelines. The diagnosis for these patients was therefore confirmed accordingto the 2011 guidelines and MDD before May 2015 [1]. Athoracic radiologist with 27 years of experience re-readthe chest high-resolution CT (HRCT) that had beenobtained at the initial evaluation. This thoracic radiologist was blind to the clinical course and examinationdata for each patient.Patients were excluded from the present study if therewas clinical evidence of other known conditions, such asconnective tissue disease, left heart failure, an occupational or environmental exposure that may result ininterstitial lung disease, or a history of ingestion of anagent known to cause interstitial lung disease. Moreover,patients who had been prescribed medication for IPF (i.e.anti-fibrotic drug, corticosteroid, immunosuppressant), orwho had undergone long-term oxygen therapy beforeinitial evaluation at a center for respiratory diseases(Aichi, Japan) were also excluded. Finally, we examined182 patients with IPF who were newly diagnosed. Ofthem, 55 patients who were in a clinical trial wereenrolled. The present study was approved by a local institutional review board (IRB No. 509).Data collectionClinical data were retrospectively collected from a medicalchart review. The eligible patients had undergone all ofPage 2 of 6the tests and assessments that were physical examinationand assessment of physiological function, dyspnoea (modified Medical Research Council (MMRC) scale andBaseline Dyspnoea Index (BDI) score), exercise capacity(6-minute-walk test (6MWT) and desaturation during6MWT), and HRQL, which was assessed using theSGRQ [6] at the initial evaluation. The Japanese versionof the SGRQ has been previously validated [20].All patients completed pulmonary function tests(PFTs) by spirometry (CHESTAC-55 V; Chest, Tokyo,Japan), according to the ATS/ERS criteria [21]. Diffusioncapacity for carbon monoxide (DLco) was also measured(CHESTAC-55 V). The values for forced vital capacity(FVC), forced expiratory volume in 1 second (FEV1),and DLco were measured according to the AmericanThoracic Society/European Respiratory Society recommendation [22]. The 6MWT was performed accordingto the ATS/ERS criteria [23]. The duration from initialevaluation to the last attendance or death was recorded.We analyzed censored cases by calling them to confirmtheir life-or-death status.Statistical analysisWe performed all analyses by using SPSS (version 22;Chicago, IL, USA). Clinical variables were used as continuous variables, except that the categorical variables ofgender, smoking status, and a history of surgical lungbiopsy were coded as one or zero for the analyses. Continuous variables were presented as mean standarddeviation unless otherwise stated. Categorical variableswere reported as counts and percentages. The survivaltime was calculated with the life table method. We performed univariate and multivariate Cox proportionalhazards analyses to investigate the relationships betweenclinical variables and mortality rate with adjustment forage and gender. Results of Cox proportional hazardsanalyses were presented in terms of estimated hazardratios (HRs) with corresponding 95% confidence intervals (CI); p values of less than 0.05 were considered tobe statistically significant. In order to select final prognostic predictors in multivariate analysis, all candidatepredictors for which the p-value was 0.1 on univariateanalysis were included in a forward selection method,with a p-value of 0.05 used for final entry or removal.To avoid multicollinearity, some of the highly correlatedvariables were excluded on multivariate analysis if theyhad Pearson’s correlation coefficients higher than 0.7.We performed a receiver operating characteristic (ROC)curve analysis to find an optimized cutoff value for 3-yearsurvival prediction. Survival curves were estimated usingthe Kaplan–Meier method, and compared using thelog-rank test across the higher and lower groups of theSGRQ total score.

Furukawa et al. Respiratory Research (2017) 18:18Page 3 of 6ResultsPatient characteristicsThe characteristics of the study population are shown inTable 1. Average age was 65.6 years old, and 85.2% ofthe patients were male. Surgical lung biopsy was performed to diagnose IPF in 97 patients (53.3%). Medianfollow-up time was 36.1 months (interquartile range,19.3 to 48.6 months), 12 cases were lost to follow-up,and 94 patients (51.6%) died within the follow-up times.MST was 48.3 months (95% CI, 43.5 to 53.1 months).Baseline levels of clinical indicesTable 1 shows all variables we collected in the presentstudy: mean levels of PaO2, the pulmonary function test,exercise capacity, dyspnoea scale and HRQL score. Meanpercent predicted FVC was 79.7%, mean percent predictedDLco was 58.2%. Mean score on the SGRQ total domainwas 34.5 points and its distribution is shown in Fig. 1. Therelationship between SGRQ and baseline physiologicalmeasures is shown in Additional file 1: Table S1. Comorbidities, especially diabetes and orthopedic disease, wereTable 1 Baseline characteristics of study populationa (N 182)CharacteristicAge, y65.6 8.0Male gender, N (%)155 (85.2)BMI, kg/m224.0 3.4Ever smoking, N (%)147 (81.9)Surgical Lung Biopsy, N (%)97 (53.3)Arterial blood gas analysisPaO2, mmHg82.0 11.6Pulmonary functionFVC, % predicted79.7 19.1FEV1 / FVC85.2 7.5DLco, % predicted (N 179)58.2 20.1Dyspnoea indexMMRC (0/1/2/3/4)51 / 74 / 41 / 15 / 1BDI (N 180)8.9 2.46MWT (N 181)6MWD, m580 136SpO2 nadir, %82.8 9.1SGRQ scoreSymptom44.2 22.6Activity40.0 26.1Impact27.7 19.8Total34.5 20.2BMI body mass index, FVC % predicted percent predicted forced vital capacity;FEV1 forced expiratory volume in the first second, DLCO % predicted percentpredicted diffusion capacity for carbon monoxide, PaO2 partial pressure foroxygen, 6MWD 6-min walk distance, SpO2 oxygen saturation by pulse oximetryaPlus–minus values are means SDFig. 1 The distribution histogram for SGRQrelated to SGRQ total score. However, they did not havepredictive value for mortality with adjustment for age andgender (HR, 1.449; 95% CI 0.901–2.332, P 0.13, HR,0.748; 95% CI 0.325–1.720, P 0.49, respectively) (Additional file 1: Table S2).Correlation between baseline levels of clinical indices andmortalityOn univariate analysis, all domain scores of the SGRQshowed significant predictive value for mortality (Table 2).Moreover, symptom, impact, and total domain showed significant predictive value for mortality with adjustment forage, gender, and %FVC model (Additional file 1: Table S3).Table 2 Univariate Cox proportional-hazard analysisVariablesAdjusted HRs (95% CI)ap valueAge, y1.005 (0.979–1.031)0.72Male gender0.725 (0.394–1.334)0.30BMI, kg/m20.882 (0.826–0.942) 0.001PaO2, mmHg0.978 (0.961–0.996)0.014FVC, % predicted0.954 (0.941–0.967) 0.001DLco, % predicted0.976 (0.964–0.989) 0.001MMRC1.413 (1.14–1.753)0.002BDI0.830 (0.764–0.903) 0.0016MWD, m0.996 (0.994–0.998) 0.001SpO2 nadir, %0.962 (0.944–0.98) 0.001SGRQaSymptom1.019 (1.01–1.029) 0.001Activity1.014 (1.006–1.022)0.001Impact1.019 (1.009–1.029) 0.001Total1.021 (1.011–1.032) 0.001; age, gender adjusted

Furukawa et al. Respiratory Research (2017) 18:18Page 4 of 6Because the SGRQ total score included all domain scoresof the SGRQ and there were high correlations between thetotal score and each domain score, we only used theSGRQ total score in multivariate analysis. Moreover, BDIwas excluded from multivariate analysis because of its highcorrelation with MMRC. In multivariate analysis using aforward selection method, the SGRQ total score was anindependent predictor for mortality (HR, 1.012; 95% CI,1.001–1.023; P .029), along with percent predicted FVC(HR, 0.957; 95% CI, 0.944–0.971; P .001) (Table 3).The ROC curve analysis for the 3-year mortality ratedemonstrated that the most accurate and optimal cutoffvalue for the SGRQ total score was 30 points (areaunder the curve, 71.1%; 95% CI 62.8–79.4%; P .001)(Fig. 2). MST was significantly shorter in patients withSGRQ total scores of over 30 points than in patientswith scores under 30 points (MST, 37.7 vs 54.7 months.HR, 2.047; 95% CI, 1.329–3.153; P .001) (Fig. 3).DiscussionThis is the first report to show that health statusassessed using the SGRQ total score is an independentpredictor of mortality in patients with IPF. Moreover,IPF patients with a SGRQ total score higher than 30points had a higher mortality rate.The SGRQ is widely used as an assessment tool forHRQL and is applied as one of the key endpoints in