All cases with histological diagnosis of interstitial pneumonia or pulmonary fibrosis involving patients of the Ruhrlandklinik Essen – West German Lung Center at the University Hospital Essen between 1993 and 2000 were retrospectively selected from the archive of the Department of Pathology at the Ruhr-University Bochum. The indication to perform open lung biopsies had been made by the clinicians on the basis of the available guidelines at that time. All patients have given written consent to surgical procedures und scientific evaluation of data. Data protection was done according to legal foundations. Patients without open lung biopsy (OLB) were excluded. Clinical records were reviewed in each case, follow up was requested by contacting local practitioners. Digital images or photographs of thoracic imaging (CXR, CT and HRCT) were only partially available for diagnostic review and written descriptions of radiological findings were often of poor quality. Therefore results of thoracic imaging were not included in the evaluation process.

BAL was performed during local anesthesia using fiberoptic bronchoscopy according to established guidelines 15 . In brief, a flexible bronchoscope was wedged into a segmental bronchus of the middle lobe or the lingula. Sterile isotonic saline was instilled in five to ten 20 ml aliquots up to a total volume of 100–200 ml, with immediate aspiration by gentle suction after each aliquot. The recovered BAL fluid was immediately processed in the laboratory. The fluid was pooled, filtered through two layers of gauze, and centrifuged at 500 g for 10 min at room temperature. The cells were counted in a haemocytometer. Slides were stained with May-Grünwald-Giemsa stain (Merck, Germany) and a total of 600 cells were counted for the cell differentials. A trypan blue exclusion test was performed for evaluating cell viability.

Slides were stained with hematoxylin and eosin in all cases, further slides with Elastica van Gieson, Periodic Acid Schiff and Prussian Blue stains were available in parts. In cases of suspected Langerhans cell histiocytosis immunohistochemistry with monoclonal antibodies against S-100 Protein and CD1a was done (ABC method, DAKO Hamburg, Germany). Additionally in one case of LIP staining for lymphocyte subtypes (CD 3 and CD 20, ABC method, DAKO Hamburg, Germany) was performed.

Diagnostic evaluation was done in three steps. Firstly, diagnosis was made as a slide review on the basis of histological patterns found by the pathologist (DT), blinded to the BAL results and other clinical data. Confidence of histological diagnosis was estimated as certain, probable or possible. If no clear cut diagnosis could be made, the most probable diagnosis was listed. Secondly, histological re-evaluation was done on the basis of clinical diagnoses made by the clinician given on the pathology request form. Thirdly, final diagnosis was made by the pulmonologist (UC) and the pathologist (DT) on the light of all clinical, radiological and histological data as interdisciplinary case evaluation (Figure 1). The results of these three diagnostic steps were compared.

Data are expressed as mean ± standard deviation. We used t-student test and ANOVA, for two or multiple groups respectively, to compare normally distributed variables. Comparison of non-normally distributed variables between two groups was done with the Mann–Whitney U test. Comparison of categorical variables between two groups, including the results of the single diagnostic steps vs interdisciplinary case evaluation, was done by the Fisher’s exact probability test. All statistical analyses were done using SPSS version 17.0 for Windows (SPSS Inc., Chicago, IL). Differences were considered statistically significant when the p value was < 0.05.

63 patients (72%) were diagnosed as idiopathic interstitial pneumonias by interdisciplinary case evaluation according to ATS/ERS criteria (Table 1). There were 27 (43%) patients with idiopathic pulmonary fibrosis (IPF), 14 (22%) with non specific interstitial pneumonia (NSIP), 6 (9.5%) with respiratory bronchiolitis with interstitial lung disease (RBILD), 3 (5%) with desquamative interstitial pneumonia (DIP), 10 (16%) with cryptogenic organizing pneumonia (COP) and 3 (5%) with lymphocytic interstitial pneumonia (LIP).

25 patients (28%) had interstitial pneumonia of non idiopathic origin. Among these were 10 cases of hypersensitivity pneumonitis (11%), 7 of Langerhans cell histiocytosis (8%), and 8 of interstitial pneumonias of other known causes and associations (9%). Non idiopathic interstitial pneumonias could not be diagnosed without interdisciplinary case evaluation (first step: 12% vs. second step: 44% correctness).

Histological patterns alone agreed with the final diagnosis in 67% of the cases. In 82% histology and clinical information could provide correct clinical diagnosis compared with final case evaluation. This provides a significant improvement of correct diagnosis, but only interdisciplinary case evaluation clarified all cases adequately. Strong distinctions are found in the different diseases.

Diagnosis of IPF/UIP based on the histological pattern alone was withdrawn in 2 cases after receiving clinical information and in further 2 cases after interdisciplinary case evaluation. UIP pattern is compatible with the clinical diagnosis of IPF if other etiological factors can be excluded. A questionnaire of 11 centres for pulmonary fibrosis published in Germany 2003, involving 62 patients revealed that IPF had been diagnosed usually 21 months after the beginning of symptoms 40 . Open lung biopsies were taken in 34%, high resolution CT was done in 71% and precipitins were analysed only in 33%. These data point to a relatively poor standard in the diagnosis of IPF and excluding interstitial pneumonias of known origin and association with UIP pattern before the publication of the ATS/ERS statement. Limitations of the clinical criteria for diagnosis of IPF have been demonstrated in other studies 41 . In HRCT honeycombing was seen in 44.4% of our cases with IPF. According to the actual guideline for IPF OLB is no longer necessary in cases with proven IPF in the appropriate clinical and radiological setting (HRCT with UIP pattern). Therefore OLB will now only be done in cases with possible IPF by history and HRCT 8 .

RBILD was overdiagnosed by histology plus clinical information compared with final evaluation (10 vs. 6 cases). A variable amount of RB can be found in smokers, which might explain overdiagnosis in those cases. Also in smokers with Langerhans cell histiocytosis RB is usually seen.

Idiopathic NSIP was overdiagnosed by histology plus clinical information compared with final evaluation (20 vs. 16 cases). Diagnosis of non-specific changes is difficult, therefore further evaluation is particularly crucial.

COP was nearly correctly diagnosed by histology plus clinical information compared with final evaluation (12 vs. 13 cases). In one case diagnosis was withdrawn in step 2 because changes were no longer evaluated as adequately expressed. The diagnosis of OP can be difficult if changes are only seen in small parts of the tissue. In this entity, correlation with radiomorphology (patchy consolidation) is essential for the correct diagnosis.

Microscopical diagnosis of DIP and LIP was done without problems and remained the final diagnosis in all steps. The DAD pattern could be diagnosed histologically as well, but proved to be associated with rheumatoid arthritis after case evaluation.

Diagnosis of non IIP without clinical information is difficult. Only 3 of 7 cases with LCH were diagnosed by histology alone. In clinically suspected LCH immunohistochemistry has to be done, which achieved correct diagnosis in all cases. In our cases HP was not suspected by histology alone. With clinical information, histological features of HP were recognized in 3 of 10 patients. In chronic HP a variety of changes ranging from UIP or NSIP patterns to OP is known 42 . In the smoking related diseases RBILD and LCH an overlap of patterns is seen regularly, which is reported in the literature as well 43 .

Interstitial pneumonias of other known causes and associations are characterized by morphological patterns also seen in IIP, the only difference is the detection of a known etiological factor 44 . Without carefully evaluated clinical history and laboratory examinations these diseases cannot be diagnosed.

With regards to the BAL findings, we found that differential cytology is of additional value in the diagnostic process. This was also shown in other studies (Table 7). Highest degree of lymphocytosis in IIP is seen in COP. A percentage of over 60% is suggestive for HP. Entities like bronchiolocentric interstitial pneumonia 45 , idiopathic bronchiolitis 46 47 48 or inhaled drug induced ILD 49 , have to be excluded histologically and clinically. Further aspects in diagnosis, prognosis and therapy of IP will be offered by systemic biology 50 , surfactant expression and immunohistochemistry 51 52 , and genetics 53 .

*mast cells are not mentioned in all studies, therefore total numbers (percentage) do not reach 100% in each study. **results were not available in all cases.

Open lung biopsies are only undertaken in otherwise unclear cases. Histology may show only non characteristic features. Case specific evaluation of clinical and pathological data is therefore a necessity for the exact diagnosis of IP, which was demonstrated by other studies 10 13 and recommended by other experts 17 . Major limitations of our study were the incompleteness of reviewable HRCT data for all patients and the lack of a second pathologist to calculate the level of agreement.