Histopathology 2014, 65, 549–560. DOI: 10.1111/his.12415
Prognostic relevance of histological variants in nonspecific interstitial pneumonia Marianne Kambouchner,1 Pierre Levy,2 Andrew G Nicholson,3 Kirsten Schubel,4 Eline Magois,5 Severine Feuillet,6 Dominique Valeyre,4,7 Jean-Francßois Bernaudin4,8 & Hilario Nunes4,7 1
Service d’Anatomie Pathologique, AP-HP, H^opital Avicenne, Bobigny, France, 2Departement de Sante Publique, AP-HP, H^opital Tenon, Universite Pierre et Marie Curie Paris 6, Paris, France, 3Department of Histopathology, Royal Brompton and Harefield NHS Foundation Trust and NHLI Division, Imperial College London, London, UK, 4Service de Pneumologie, AP-HP, H^opital Avicenne, Bobigny, France, 5Service de Pneumologie, H^opital d’Amiens, Universite de Picardie Jules Verne, Amiens, France, 6Service de Pneumologie, AP-HP, H^opital Saint-Louis, H^opital Saint Louis, Universite Paris Diderot, Sorbonne Paris Cite, Paris, France, 7Universite Paris 13, Sorbonne Paris Cite, Paris, France, and 8Service d’ Anatomie Pathologique, AP-HP, H^opital Tenon, Universite Pierre et Marie Curie Paris 6, Paris, France Date of submission 9 January 2014 Accepted for publication 8 March 2014 Published online Article Accepted 8 March 2014
Kambouchner M, Levy P, Nicholson A G, Schubel K, Magois E, Feuillet S, Valeyre D, Bernaudin J-F & Nunes H (2014) Histopathology 65, 549–560
Prognostic relevance of histological variants in nonspecific interstitial pneumonia Aims: Although histological non-specific interstitial pneumonia (NSIP) is concisely defined, overlap with other patterns is described. While most frequently idiopathic, NSIP is seen in various clinical contexts such as connective tissue diseases (CTDs) and chronic hypersensitivity pneumonitis (cHP). This study was designed to determine if NSIP could be separated into subgroups based on minor histological features and to correlate these subgroups with clinical associations and survival. Methods and results: One hundred and thirty-six patients with biopsy-proven NSIP were included [clinical diagnosis: CTDs (23%), cHP (12%), idiopathic (65%)]. In addition to the agreed NSIP criteria, seven subgroups were identified: essential NSIP and six overlap subgroups according to superimposed minor histological features. Interobserver concordance resulted in the following consensus: essential NSIP (36%), usual interstitial pneumonia (UIP) overlap (26%), cHP over-
lap (10%), organizing pneumonia (OP) overlap (6%), organizing diffuse alveolar damage (DAD) overlap (10%), desquamative interstitial pneumonia overlap (7%) and lymphoid interstitial pneumonia overlap (2%). OP overlap was associated with CTDs (P = 0.04) and cHP overlap with a cHP clinical diagnosis (P = 0.02). Survival was different between subgroups (P = 0.0002). Organizing DAD overlap exhibited poorer survival at 5 years (32%), followed by UIP overlap (57%). Independent predictors of mortality were organizing DAD overlap (HR = 4.99, 95% CI = 2.15– 11.58, P = 0.0002), UIP overlap (HR = 2.11, 95% CI = 1.12–3.99, P = 0.02) and a clinical diagnosis of cHP (HR = 2.17, 95% CI = 1.05–4.47, P = 0.035). Conclusions: Non-specific interstitial pneumonia subdivision into pathological subgroups is clinically relevant from a prognostic and causal perspective.
Keywords: aetiology, non-specific interstitial pneumonia, pathology, survival
Introduction Address for correspondence: Dr M Kambouchner, Service de d’Anatomie Pathologique, H^ opital Avicenne, 125 rue de Stalingrad, 93009 Bobigny, France. e-mail: [email protected] © 2014 John Wiley & Sons Ltd.
The pathological classification of idiopathic interstitial pneumonias (IIPs) has evolved considerably over the past 20 years, since Katzenstein and Fiorelli proposed
550
M Kambouchner et al.
the term ‘non-specific interstitial pneumonia’ (NSIP) in 1994 to refer to cases that could not be classified readily into one of the main patterns recognized in the previous Liebow classification.1,2 Subsequently, published data have shown that NSIP is not simply a ‘wastebasket’ group and that its differentiation from usual interstitial pneumonia (UIP) carries major prognostic implications.1,3–7 In the 2002 ATS/ERS consensus committee classification of the IIPs, NSIP was recognized as a specific histological pattern,8 and in the 2013 update, idiopathic NSIP was accepted as a distinct clinical entity, while the NSIP pattern was recognized to be observed in a variety of settings.9 The NSIP histological pattern comprises varying degrees of alveolar wall inflammation or fibrosis, the key feature being the temporal uniformity of lesions, which contrasts with the heterogeneous/patchwork involvement observed in UIP.8–11 Although considered as pertinent negative findings, several minor features are occasionally present and do not refute the pathological diagnosis of NSIP [non-prominent organizing pneumonia (OP), inconspicuous fibroblastic foci and granulomas, some degree of subpleural and paraseptal distribution, and peribronchiolar distribution].8–11 While most frequently idiopathic, the NSIP histological pattern can also be related to diverse causes, including connective tissue diseases (CTDs), chronic hypersensitivity pneumonitis (cHP),12 drug toxicity and slowly resolving diffuse alveolar damage (DAD).1 In this context, several authors have set up a schematic representation of the NSIP pathological pattern with an overlap with other interstitial lung diseases, including UIP, cHP, OP, DAD, desquamative interstitial pneumonia (DIP) and lymphoid interstitial pneumonia (LIP) (Figure 1).13 Accordingly, we hypothesized that the minor histological features might be the framework for refining the NSIP histological pattern into subgroups. The aim of our study was therefore to determine whether or not these subgroups are associated with the underlying causes and/or with a distinct prognosis.
Methods PATIENT SELECTION AND BIOPSY TECHNIQUE
This retrospective study received institutional review board approval (Comite de Protection des Personnes Ile-de-France, no. 2012-12-01). All 136 consecutive patients with a pathological diagnosis of NSIP on surgical lung biopsies (SLB) (either open thoracotomy or
Figure 1. Overlaps of NSIP with other pathological patterns of interstitial lung diseases (adapted from Travis et al.13). NSIP, nonspecific interstitial pneumonia; UIP, usual interstitial pneumonia; cHP, chronic hypersensitivity pneumonitis; OP, organizing pneumonia; DAD, diffuse alveolar damage; DIP, desquamative interstitial pneumonia; LIP, lymphocytic interstitial pneumonia.
video-assisted thoracoscopy) seen at the Pathology Department of Avicenne University Hospital from 1983 to 2010 were included. The pathological diagnosis, based on the ATS/ERS international classification of the IIPs,8,9 was consensually approved by two expert pathologists (A.G.N. and M.K.). Biopsy specimens were obtained from one single lobe in 32 cases (23.5%), two lobes in 76 cases (55.9%) and three lobes in four cases (2.9%). For 24 patients (17.6%), their localization was not available. For the 112 others the sites were: upper lobes: n = 78 (69.6%), lower lobes: n = 92 (82.1%), middle lobe: n = 19 (16.9%) or lingula: n = 5 (4.5%). The analysed tissue surface per patient ranged from 0.6 to 22.7 cm2, with a mean of 5.6 3.9 cm2 and a median of 4.7 cm2. PATHOLOGICAL REVIEW
Biopsies were analysed by two pathologists (M.K. and J.F.B.) in three steps. First, seven pathological NSIP subgroups were defined: i.e. essential NSIP and six overlap patterns. Secondly, the series was reviewed independently by M.K. and J.F.B., who were blinded to clinico-radiological data. A unique pattern was attributed to each patient, which resulted from the examination of all available slides. Thirdly, a consensus was found in discordant cases. © 2014 John Wiley & Sons Ltd, Histopathology, 65, 549–560.
Non-specific interstitial pneumonia variants
551
DEFINITION OF NSIP SUBGROUPS
STATISTICAL ANALYSIS
In addition to the typical major pathological features of NSIP, superimposed minor pathological features were the cornerstone for the definition of subsets of NSIP overlap patterns. Essential NSIP referred to cases with an absence of superimposed minor pathological features. The NSIP/UIP overlap (Figure 2) is defined by topographic similarities with UIP, but with no evidence of clear temporal heterogeneity, no or very rare fibroblastic foci and variable chronic inflammation compared to UIP. However, the fibrotic process is relatively patchy, with a reinforced subpleural and paraseptal distribution. When abnormal and subnormal lung areas coexist, fibrosis blends into the normal areas rather than being sharply demarcated without transition. In the NSIP/cHP overlap (Figure 2), the fibroinflammatory process has focal peribronchiolar accentuation. Although inconspicuous or absent, rare ill-defined non-necrotizing granulomas can be observed. The NSIP/OP overlap (Figure 2) is characterized by the presence of scarce intraluminal fibro-inflammatory buds in distal airspaces. By contrast with OP, there is significant interstitial fibrosis and lung architecture is not preserved. In the NSIP/organizing DAD overlap (Figure 3), there are numerous areas of fibroblastic proliferation in a myxoid stroma resulting in the thickening of alveolar septa. Collagen deposition is relatively mild and inflammation is variable. There is no acute lung injury pattern; in particular, hyaline membranes and fibrin deposits are absent. Regarding NSIP/DIP overlap (Figure 3), although the accumulation of alveolar macrophages is not prominent, this term was chosen because of architectural similarities with DIP.8,10,14 Interstitial changes are moderate, consisting of a uniform thickening of the alveolar septa and minimal collagen deposition. In the NSIP/LIP overlap, there is extensive alveolar septal infiltration by lymphocytes and/or plasma cells. By contrast with LIP, there is a significant interstitial fibrosis and lung architecture is not preserved.
Results are expressed as percentages or means standard deviations. Interobserver concordance was measured with Cohen’s kappa coefficient. The various pathological groups were compared using Fisher’s exact test and the Kruskal–Wallis test for categorical and continuous variables, respectively. When the Kruskal–Wallis test indicated that groups differed, posthoc comparisons were performed on the relevant continuous measures using the Mann–Whitney U-test. Survival was calculated from the date of SLB until the end of follow-up. The probability of survival of groups was estimated using the Kaplan–Meier method and compared by log-rank test. Patients not classified or with NSIP/LIP overlap were not included in survival analysis because of the small number of cases. Univariate analysis used the log-rank test. For continuous variables, we chose to divide patients into two groups on both sides of the median value. Parameters with a P-value
Prognostic relevance of histological variants in nonspecific interstitial pneumonia Marianne Kambouchner,1 Pierre Levy,2 Andrew G Nicholson,3 Kirsten Schubel,4 Eline Magois,5 Severine Feuillet,6 Dominique Valeyre,4,7 Jean-Francßois Bernaudin4,8 & Hilario Nunes4,7 1
Service d’Anatomie Pathologique, AP-HP, H^opital Avicenne, Bobigny, France, 2Departement de Sante Publique, AP-HP, H^opital Tenon, Universite Pierre et Marie Curie Paris 6, Paris, France, 3Department of Histopathology, Royal Brompton and Harefield NHS Foundation Trust and NHLI Division, Imperial College London, London, UK, 4Service de Pneumologie, AP-HP, H^opital Avicenne, Bobigny, France, 5Service de Pneumologie, H^opital d’Amiens, Universite de Picardie Jules Verne, Amiens, France, 6Service de Pneumologie, AP-HP, H^opital Saint-Louis, H^opital Saint Louis, Universite Paris Diderot, Sorbonne Paris Cite, Paris, France, 7Universite Paris 13, Sorbonne Paris Cite, Paris, France, and 8Service d’ Anatomie Pathologique, AP-HP, H^opital Tenon, Universite Pierre et Marie Curie Paris 6, Paris, France Date of submission 9 January 2014 Accepted for publication 8 March 2014 Published online Article Accepted 8 March 2014
Kambouchner M, Levy P, Nicholson A G, Schubel K, Magois E, Feuillet S, Valeyre D, Bernaudin J-F & Nunes H (2014) Histopathology 65, 549–560
Prognostic relevance of histological variants in nonspecific interstitial pneumonia Aims: Although histological non-specific interstitial pneumonia (NSIP) is concisely defined, overlap with other patterns is described. While most frequently idiopathic, NSIP is seen in various clinical contexts such as connective tissue diseases (CTDs) and chronic hypersensitivity pneumonitis (cHP). This study was designed to determine if NSIP could be separated into subgroups based on minor histological features and to correlate these subgroups with clinical associations and survival. Methods and results: One hundred and thirty-six patients with biopsy-proven NSIP were included [clinical diagnosis: CTDs (23%), cHP (12%), idiopathic (65%)]. In addition to the agreed NSIP criteria, seven subgroups were identified: essential NSIP and six overlap subgroups according to superimposed minor histological features. Interobserver concordance resulted in the following consensus: essential NSIP (36%), usual interstitial pneumonia (UIP) overlap (26%), cHP over-
lap (10%), organizing pneumonia (OP) overlap (6%), organizing diffuse alveolar damage (DAD) overlap (10%), desquamative interstitial pneumonia overlap (7%) and lymphoid interstitial pneumonia overlap (2%). OP overlap was associated with CTDs (P = 0.04) and cHP overlap with a cHP clinical diagnosis (P = 0.02). Survival was different between subgroups (P = 0.0002). Organizing DAD overlap exhibited poorer survival at 5 years (32%), followed by UIP overlap (57%). Independent predictors of mortality were organizing DAD overlap (HR = 4.99, 95% CI = 2.15– 11.58, P = 0.0002), UIP overlap (HR = 2.11, 95% CI = 1.12–3.99, P = 0.02) and a clinical diagnosis of cHP (HR = 2.17, 95% CI = 1.05–4.47, P = 0.035). Conclusions: Non-specific interstitial pneumonia subdivision into pathological subgroups is clinically relevant from a prognostic and causal perspective.
Keywords: aetiology, non-specific interstitial pneumonia, pathology, survival
Introduction Address for correspondence: Dr M Kambouchner, Service de d’Anatomie Pathologique, H^ opital Avicenne, 125 rue de Stalingrad, 93009 Bobigny, France. e-mail: [email protected] © 2014 John Wiley & Sons Ltd.
The pathological classification of idiopathic interstitial pneumonias (IIPs) has evolved considerably over the past 20 years, since Katzenstein and Fiorelli proposed
550
M Kambouchner et al.
the term ‘non-specific interstitial pneumonia’ (NSIP) in 1994 to refer to cases that could not be classified readily into one of the main patterns recognized in the previous Liebow classification.1,2 Subsequently, published data have shown that NSIP is not simply a ‘wastebasket’ group and that its differentiation from usual interstitial pneumonia (UIP) carries major prognostic implications.1,3–7 In the 2002 ATS/ERS consensus committee classification of the IIPs, NSIP was recognized as a specific histological pattern,8 and in the 2013 update, idiopathic NSIP was accepted as a distinct clinical entity, while the NSIP pattern was recognized to be observed in a variety of settings.9 The NSIP histological pattern comprises varying degrees of alveolar wall inflammation or fibrosis, the key feature being the temporal uniformity of lesions, which contrasts with the heterogeneous/patchwork involvement observed in UIP.8–11 Although considered as pertinent negative findings, several minor features are occasionally present and do not refute the pathological diagnosis of NSIP [non-prominent organizing pneumonia (OP), inconspicuous fibroblastic foci and granulomas, some degree of subpleural and paraseptal distribution, and peribronchiolar distribution].8–11 While most frequently idiopathic, the NSIP histological pattern can also be related to diverse causes, including connective tissue diseases (CTDs), chronic hypersensitivity pneumonitis (cHP),12 drug toxicity and slowly resolving diffuse alveolar damage (DAD).1 In this context, several authors have set up a schematic representation of the NSIP pathological pattern with an overlap with other interstitial lung diseases, including UIP, cHP, OP, DAD, desquamative interstitial pneumonia (DIP) and lymphoid interstitial pneumonia (LIP) (Figure 1).13 Accordingly, we hypothesized that the minor histological features might be the framework for refining the NSIP histological pattern into subgroups. The aim of our study was therefore to determine whether or not these subgroups are associated with the underlying causes and/or with a distinct prognosis.
Methods PATIENT SELECTION AND BIOPSY TECHNIQUE
This retrospective study received institutional review board approval (Comite de Protection des Personnes Ile-de-France, no. 2012-12-01). All 136 consecutive patients with a pathological diagnosis of NSIP on surgical lung biopsies (SLB) (either open thoracotomy or
Figure 1. Overlaps of NSIP with other pathological patterns of interstitial lung diseases (adapted from Travis et al.13). NSIP, nonspecific interstitial pneumonia; UIP, usual interstitial pneumonia; cHP, chronic hypersensitivity pneumonitis; OP, organizing pneumonia; DAD, diffuse alveolar damage; DIP, desquamative interstitial pneumonia; LIP, lymphocytic interstitial pneumonia.
video-assisted thoracoscopy) seen at the Pathology Department of Avicenne University Hospital from 1983 to 2010 were included. The pathological diagnosis, based on the ATS/ERS international classification of the IIPs,8,9 was consensually approved by two expert pathologists (A.G.N. and M.K.). Biopsy specimens were obtained from one single lobe in 32 cases (23.5%), two lobes in 76 cases (55.9%) and three lobes in four cases (2.9%). For 24 patients (17.6%), their localization was not available. For the 112 others the sites were: upper lobes: n = 78 (69.6%), lower lobes: n = 92 (82.1%), middle lobe: n = 19 (16.9%) or lingula: n = 5 (4.5%). The analysed tissue surface per patient ranged from 0.6 to 22.7 cm2, with a mean of 5.6 3.9 cm2 and a median of 4.7 cm2. PATHOLOGICAL REVIEW
Biopsies were analysed by two pathologists (M.K. and J.F.B.) in three steps. First, seven pathological NSIP subgroups were defined: i.e. essential NSIP and six overlap patterns. Secondly, the series was reviewed independently by M.K. and J.F.B., who were blinded to clinico-radiological data. A unique pattern was attributed to each patient, which resulted from the examination of all available slides. Thirdly, a consensus was found in discordant cases. © 2014 John Wiley & Sons Ltd, Histopathology, 65, 549–560.
Non-specific interstitial pneumonia variants
551
DEFINITION OF NSIP SUBGROUPS
STATISTICAL ANALYSIS
In addition to the typical major pathological features of NSIP, superimposed minor pathological features were the cornerstone for the definition of subsets of NSIP overlap patterns. Essential NSIP referred to cases with an absence of superimposed minor pathological features. The NSIP/UIP overlap (Figure 2) is defined by topographic similarities with UIP, but with no evidence of clear temporal heterogeneity, no or very rare fibroblastic foci and variable chronic inflammation compared to UIP. However, the fibrotic process is relatively patchy, with a reinforced subpleural and paraseptal distribution. When abnormal and subnormal lung areas coexist, fibrosis blends into the normal areas rather than being sharply demarcated without transition. In the NSIP/cHP overlap (Figure 2), the fibroinflammatory process has focal peribronchiolar accentuation. Although inconspicuous or absent, rare ill-defined non-necrotizing granulomas can be observed. The NSIP/OP overlap (Figure 2) is characterized by the presence of scarce intraluminal fibro-inflammatory buds in distal airspaces. By contrast with OP, there is significant interstitial fibrosis and lung architecture is not preserved. In the NSIP/organizing DAD overlap (Figure 3), there are numerous areas of fibroblastic proliferation in a myxoid stroma resulting in the thickening of alveolar septa. Collagen deposition is relatively mild and inflammation is variable. There is no acute lung injury pattern; in particular, hyaline membranes and fibrin deposits are absent. Regarding NSIP/DIP overlap (Figure 3), although the accumulation of alveolar macrophages is not prominent, this term was chosen because of architectural similarities with DIP.8,10,14 Interstitial changes are moderate, consisting of a uniform thickening of the alveolar septa and minimal collagen deposition. In the NSIP/LIP overlap, there is extensive alveolar septal infiltration by lymphocytes and/or plasma cells. By contrast with LIP, there is a significant interstitial fibrosis and lung architecture is not preserved.
Results are expressed as percentages or means standard deviations. Interobserver concordance was measured with Cohen’s kappa coefficient. The various pathological groups were compared using Fisher’s exact test and the Kruskal–Wallis test for categorical and continuous variables, respectively. When the Kruskal–Wallis test indicated that groups differed, posthoc comparisons were performed on the relevant continuous measures using the Mann–Whitney U-test. Survival was calculated from the date of SLB until the end of follow-up. The probability of survival of groups was estimated using the Kaplan–Meier method and compared by log-rank test. Patients not classified or with NSIP/LIP overlap were not included in survival analysis because of the small number of cases. Univariate analysis used the log-rank test. For continuous variables, we chose to divide patients into two groups on both sides of the median value. Parameters with a P-value
