Diagnostic Value of the CD103+CD4+/CD4+ Ratio to Differentiate Sarcoidosis from Other Causes of Lymphocytic Alveolitis

Sarcoidosis is a chronic multisystemic disease of unknown origin characterized by granulomatous inflammation in various organs, which affects the lungs in 90% of cases [1]. The diagnosis of pulmonary sarcoidosis is based on clinical symptoms, chest imaging, cytologic analysis of bronchoalveolar lavage fluid (BALF), and histopathology obtained by either bronchial biopsy, transbronchial lung biopsy, transbronchial needle aspiration of mediastinal lymph nodes, or rarely surgical lung biopsy [2,3,4].

Pulmonary sarcoidosis is characterized by increased T cells in BALF compared to healthy subjects [5]. These T cells are predominantly CD4+, with a CD4+/CD8+ ratio of typically 3:1 to 5:1, as compared to 2:1 in healthy subjects [6]. However, the diagnostic value of BALF in sarcoidosis is limited by the lack of a specific and sensitive marker. BALF lymphocytosis is not specific to sarcoidosis, and the use of the CD4+/CD8+ ratio is controversial, except when it is >3.5 and only if the disease is active [7].

Integrin αEβ7 (CD103) is an adhesion molecule expressed on 95% of intraepithelial lymphocytes within mucosal tissues, but on <2% of circulating peripheral blood lymphocytes [8,9,10,11,12]. A distinct expression pattern of CD103 on BALF CD4+ lymphocytes has been observed in various interstitial lung diseases (ILDs), including sarcoidosis [12,13]. It has been hypothesized that most CD4+ lymphocytes in pulmonary sarcoidosis do not originate from bronchial mucosa-associated lymphoid tissue, but from redistribution from the peripheral blood and compartmentalization to the lung [14]. As the peripheral blood CD4+ lymphocytes do not express CD103, differential expression of this marker on BALF CD4+ lymphocytes could serve as a diagnostic marker of sarcoidosis as compared to other ILDs. Several retrospective studies have explored this issue [13,15,16,17,18]. Kolopp-Sarda et al. [15] found that a high CD4+/CD8+ ratio combined with a low CD103+CD4+/CD4+ ratio allowed to distinguish sarcoidosis from other ILDs with a 96% sensitivity. Similarly, Mota et al. [17] found a 98% sensitivity. However, two other studies showed a lower performance of the CD103+CD4+/CD4+ ratio as a diagnostic marker for sarcoidosis [16,18]. The goal of the present study was to further investigate the value of the CD103+CD4+/CD4+ ratio to differentiate sarcoidosis from other causes of lymphocytic alveolitis in a large retrospective case series.

Since 2006, flow cytometry has been routinely performed at the Institute of Pathology of Lausanne University Hospital on all BALF samples with >20% lymphocytes on differential cell count, and on samples with <20% lymphocytes in case of immunosuppression or hematological malignancies. For the purpose of this study, flow cytometry results of BALF from Lausanne University Hospital were retrospectively reviewed. Cases with mention of poor BALF quality were discarded. For the remaining cases, electronic medical records were reviewed for diagnostic information, using the pre-defined diagnostic criteria mentioned below.

The diagnosis of sarcoidosis required compatible clinical and imaging features and the presence of non-caseating granuloma on tissue biopsy or transbronchial needle aspiration cytology of mediastinal lymph nodes. Typical stage 1 sarcoidosis occurring in the context of Löfgren syndrome without biopsy or cytology was also accepted. The diagnosis of pulmonary tuberculosis required the presence of acid-fast bacilli, positive polymerase chain reaction and/or positive cultures for Mycobacterium tuberculosis on sputum, bronchial aspiration, or BALF. A diagnosis of infection other than tuberculosis required either identification of a virus or bacterium on sputum, bronchial aspiration or BALF by polymerase chain reaction and/or positive cultures, or marked clinical improvement with antibiotic therapy. Hypersensitivity pneumonitis (HP) was diagnosed according to the criteria of Schuyler and Cormier [19]. The diagnosis of drug-induced pneumonia was based on a history of exposure to a drug with known pulmonary toxicity, a temporal relationship between exposure and onset of symptoms, and improvement after drug withdrawal. Organising pneumonia (OP) was diagnosed on the basis of compatible clinical and imaging features, and presence of buds of granulation tissue on transbronchial lung biopsy without any feature suggesting another diagnosis. The diagnosis of OP was also accepted in the absence of biopsy if radiation therapy for breast cancer, a classical cause of OP, had occurred within the past year. Non-specific interstitial pneumonia (NSIP) was diagnosed on surgical lung biopsy. A diagnosis of NSIP pattern was also accepted if the clinical and imaging findings were highly suggestive even in the absence of surgical lung biopsy. Cases not meeting the above diagnostic criteria were excluded.

Included cases were grouped into nine diagnostic categories: sarcoidosis, tuberculosis, non-tuberculous infections, HP, NSIP, OP, drug-induced lung diseases, other ILDs, and other diagnoses. As some categories had a small number of cases, cases were further grouped into four larger categories: sarcoidosis, infections (including tuberculosis and non-tuberculous infections), ILDs (including HP, NSIP, OP, drug-induced lung diseases, and other ILDs), and other diagnoses. Statistical analysis was performed for both the nine- and four-category distributions.

BALF was collected in sterile bottles during regular diagnostic work-up and filtered through gauze. Ten milliliters of the sample were cytocentrifuged, then smeared and stained with May-Grünwald Giemsa, modified Papanicolaou, Prussian Blue, Gram-Weigert and Ziehl-Neelsen as needed. For differential cell count, a sample corresponding to 250,000-300,000 cells was passed through a multipore filter. After staining with modified Papanicolaou, a minimum of 2 × 100 cells were counted.

For flow cytometry, 10-30 ml of native sample were centrifuged, decanted and resuspended in NH₄Cl or Roswell Park Memorial Institute medium several times until reaching a cellularity of 2 × 10⁶. The cells were stained with fluorochrome-labeled monoclonal antibodies (Dako Denmark A/S, Glostrup, Denmark) for CD45 (FITC, clone T29/33), CD4 (RPE, clone MT310), CD8 (FITC, clone DK25), and CD103 (FITC, clone Ber-ACT8) following a standard protocol, and five-color flow cytometric analysis was performed (Navios Beckman Coulter flow cytometer).

Data analyses were performed using Stata Statistical Software, release 13 (StataCorp LP, College Station, Tex., USA). Data were summarized as median and interquartile range (IQR) for continuous variables, and as number and percentage for categorical variables. Group comparison was performed using the Wilcoxon test, and the Kruskal-Wallis test when comparing more than two groups. To assess the discriminative power of the cellular marker ratios CD103+CD4+/CD4+ and CD4+/CD8+ or their combination to distinguish sarcoidosis from other diagnoses, a ROC analysis was used. Sensitivity, specificity and precision were reported, and the best threshold for optimal sensitivity and specificity was calculated. The study was approved by the ethics committee of the canton of Vaud (CER-VD 241/13).

The case selection process is summarized in figure 1. Three hundred and ninety-one consecutive flow cytometries of BALF performed at Lausanne University Hospital between January 1st, 2006 and July 18th, 2014 were reviewed. Ninety-seven cases were excluded due to insufficient technical quality of the BALF sample. Another 87 cases were excluded because the final diagnosis was unclear or unavailable (n = 46), or because pre-defined diagnostic criteria were not met (n = 41). These 41 cases were the following: suspicion of sarcoidosis without histological diagnosis (n = 13), suspicion of tuberculosis without microbiological evidence of M. tuberculosis infection (n = 7), suspicion of non-tuberculous infection without sufficient evidence to support this diagnosis (n = 10), and suspicion of OP without histological proof (n = 11).

The remaining 207 cases fulfilling the pre-defined diagnostic criteria were included. Their distribution into nine diagnostic categories is shown in figure 1. Sarcoidosis (n = 53) was diagnosed by tissue biopsy (n = 22), transbronchial needle aspiration of mediastinal lymph nodes (n = 25), or typical Löfgren syndrome (n = 6). Non-tuberculous pulmonary infection (n = 38) was diagnosed by identification of a virus or bacterium (n = 23) or marked clinical improvement with antibiotic therapy (n = 15). OP (n = 9) was diagnosed by transbronchial biopsy (n = 5) or a history of radiation therapy to the breast (n = 4). NSIP (n = 14) was diagnosed by surgical lung biopsy (n = 2) or highly suggestive clinical and imaging features (n = 12).

The main characteristics of the study population are presented in table 1. The median age was 56 years (IQR = 31), and 100 patients (48%) were women. Patients with sarcoidosis, HP and tuberculosis were significantly younger than patients of other diagnostic categories (p = 0.0001). BALF neutrophils were significantly lower in sarcoidosis and higher in HP and non-tuberculous infections (p = 0.0001). No significant differences were found between categories for BALF lymphocyte count.

The 53 patients with sarcoidosis had a median age of 46 years (IQR = 20), and 23 (43%) were women. Their main characteristics are presented in table 2. At the time of BALF sampling, only 5 were under therapy for sarcoidosis. The median follow-up duration was 2.3 years (IQR = 3.4).

The results of flow cytometry in the nine-category distribution are presented in table 3. The median value of the CD4+/CD8+ ratio was significantly higher in sarcoidosis compared to all other diagnoses, except for tuberculosis (fig. 2a). ROC curve analysis to discriminate between sarcoidosis and all other diagnoses showed an area under the ROC curve (AUC) of 79% (fig. 3a). The best cut-off value of CD4+/CD8+ ratio for an optimal specificity and sensitivity was 3.4.

The CD103+CD4+/CD4+ ratio was significantly lower in sarcoidosis compared to NSIP, HP and other ILDs (table 3, fig. 2b). However, its discriminative power to distinguish between sarcoidosis and all other diagnoses was low, as calculated by the ROC curve analysis (AUC = 62%) (fig. 3b). The combination of CD4+/CD8+ and CD103+CD4+/CD4+ ratios resulted in an AUC of 76% (fig. 3c), which was lower than the AUC of the CD4+/CD8+ ratio alone.

The data were also analyzed after grouping cases into four large categories: sarcoidosis, all infections, all ILDs, and other diagnoses (table 4). The CD4+/CD8+ ratio was significantly higher in sarcoidosis compared to each other category individually (fig. 4a). The only significant difference in the CD103+CD4+/CD4+ ratio was between sarcoidosis and ILDs (fig. 4b), with a higher ratio in ILDs (0.14 vs. 0.08, p < 0.001).

Discrimination between sarcoidosis and all ILDs based on the CD4+/CD8+ ratio level had an AUC of 79%, with an optimal specificity (69%) and sensitivity (70%) reached at a threshold of 3.4. However, the calculated AUC for the CD103+CD4+/CD4+ ratio was only 69%. The combination of the CD4+/CD8+ and CD103+CD4+/CD4+ ratios resulted in an AUC of 78%, which was lower than that of the CD4+/CD8+ ratio alone (data not shown).

Four studies analyzing the CD103+CD4+/CD4+ ratio in sarcoidosis, either alone or in combination with the CD4+/CD8+ ratio, have been previously performed [15,16,17,18]. The cut-off values defined in those studies to discriminate between sarcoidosis and other diagnoses were applied to our study population for comparison (table 5). Sensitivity varied between 55 and 98%, and specificity between 9 and 86%. The precision of classification varied between 32 and 78%, and the AUC varied between 54 and 73%.

No correlation was found between the CD103+CD4+/CD4+ ratio and the percentage of BALF lymphocytes, neutrophils and eosinophils, neither in patients with sarcoidosis not in the whole study population (data not shown). Among patients with sarcoidosis, the CD103+CD4+/CD4+ ratio was significantly lower in those with initial radiographic stage 1 (median 0.06, IQR = 0.15, n = 28) compared to stage 2 (median 0.17, IQR = 0.17, n = 21, p = 0.02). No significant correlation was found between the CD103+CD4+/CD4+ ratio and initial values of forced expiratory volume in 1 s, forced vital capacity (FVC), total lung capacity (TLC), residual volume, and carbon monoxide transfer factor (DLCO). A significant negative correlation was found between the CD103+CD4+/CD4+ ratio and the values at last visit of FVC (n = 37, r = -0.42, p = 0.009), TLC (n = 31, r = -0.63, p = 0.0001) and DLCO (n = 34, r = -0.37, p = 0.03). The occurrence of any extrathoracic organ involvement, the use of therapy at the time of bronchoalveolar lavage, the need for therapy after diagnosis, and the sarcoidosis clinical activity classification according to Prasse et al. [20] were not correlated with the CD103+CD4+/CD4+ ratio (data not shown).

In the present study, we demonstrated that the BALF lymphocyte CD103+CD4+/CD4+ ratio does not accurately discriminate between sarcoidosis and other causes of lymphocytic alveolitis, neither alone nor in combination with the CD4+/CD8+ ratio, and is therefore not a relevant marker for the clinical diagnosis of sarcoidosis. Furthermore, the CD103+CD4+/CD4+ ratio does not accurately discriminate between sarcoidosis and other ILDs. When applying the previously published CD103+CD4+/CD4+ ratio cut-offs to our population, we generally found lower sensitivity and specificity than in the original studies.

The CD4+/CD8+ ratio has been previously investigated in order to determine whether it could provide an accurate diagnosis of sarcoidosis without lung biopsy [6,21,22,23]. In agreement with these data, we found a significant correlation between sarcoidosis and an elevated CD4+/CD8+ ratio (>3.4 in our study). However, the AUC of the CD4+/CD8+ ratio to distinguish sarcoidosis from other causes of lymphocytic alveolitis was only 79%, and it was also 79% to separate sarcoidosis from other ILDs. Thus, on its own, this marker cannot differentiate between sarcoidosis and other ILDs with enough accuracy to be useful in daily practice.

Four previous studies have investigated the CD103 integrin on CD4+ T lymphocytes in BALF as a putative marker for sarcoidosis [15,16,17,18], and several of them have suggested that a high CD4+/CD8+ ratio combined with a low CD103+CD4+/CD4+ ratio would allow to distinguish sarcoidosis from other causes of alveolar lymphocytosis. Kolopp-Sarda et al. [15] first reported that the combination of a CD4+/CD8+ ratio >2.5 and a CD103+CD4+/CD4+ ratio <0.31 had a sensitivity of 96% to discriminate between sarcoidosis and other causes of lymphocytic alveolitis. They concluded that the combination of both ratios would constitute a good marker for the diagnosis of sarcoidosis. Heron et al. [16] and Mota et al. [17] also reported a good discriminative power of the CD103+CD4+/CD4+ ratio (table 5). However, in another study by Hyldgaard et al. [18], the combination of CD4+/CD8+ and CD103+CD4+/CD4+ ratios had a sensitivity of only 42% and a specificity of 91%, which led to conclude that the combination of these ratios was a poor marker for sarcoidosis. Of note, all those studies had small study populations (table 5), and two of them included less than 20 cases with sarcoidosis [15,18].

In our population, the CD103+CD4+/CD4+ ratio had a weak discriminative power for the diagnosis of sarcoidosis, with an AUC of only 62% to distinguish sarcoidosis from all other causes of lymphocytic alveolitis, and an AUC of 69% to distinguish sarcoidosis from other ILDs. The use of combined CD4+/CD8+ and CD103+CD4+/CD4+ ratios resulted in modest increases of the AUC to 76% for sarcoidosis versus all causes of lymphocytic alveolitis, and to 78% for sarcoidosis versus other ILDs. This remained lower than the AUC of the CD4+/CD8+ ratio alone (79%), suggesting that the contribution of the CD103+CD4+/CD4+ ratio itself was low.

We also tested the cut-off values suggested by other studies on our population (table 5). The highest percentage of correctly classified patients (78%) was achieved by using Hyldgaard et al.'s [18] cut-off, whereas with the one suggested by Mota et al. [17], the percentage of correctly classified patients was very low (32%). Further analysis of our population using the CD103+CD4+/CD4+ ratio alone showed a high sensitivity but a very low specificity, and the precision did not exceed 43% (table 5). Altogether, this reinforces our own observations suggesting that the CD103+CD4+/CD4+ ratio, either alone or combined with the CD4+/CD8+ ratio, does not provide sufficient diagnostic precision to identify sarcoidosis among other causes of lymphocytic alveolitis.

The strength of our study was the size of the population (n = 207), which was much larger than in any other study of the CD103+CD4+/CD4+ ratio. The number of sarcoidosis cases (n = 53) was also higher than in most other studies and almost equaled the 56 cases reported by Heron et al. [16]. Unlike the other studies, we were able to precisely classify our population into well-defined diagnostic categories and to compare sarcoidosis to each category individually, as well as to broader diagnostic groups. This allowed a more precise analysis of the CD103+CD4+/CD4+ ratio for other diagnoses such as HP, NSIP, infections, and other ILDs.

One limitation of our study is the fact that we included only patients with BALF lymphocytosis >20%, whereas there have been reports of cases of sarcoidosis with normal or only mildly elevated lymphocyte counts [18,21,24,25], especially in asymptomatic patients [26]. Although this occurs in a minority of sarcoidosis patients, this population would have been missed by our study design, and we therefore cannot rule out that the CD103+CD4+/CD4+ ratio might have greater relevance in sarcoidosis with mild or no BALF lymphocytosis. Another limitation is that not all diagnostic categories were equally represented. For instance, there was only a small number of patients with HP in our population (n = 6). This prevented us from directly comparing sarcoidosis to HP in order to explore the value of the CD103+CD4+/CD4+ ratio in separating these two entities. One recent study found a significantly higher CD103+CD4+/CD4+ ratio in HP as compared to sarcoidosis, but no cut-off value was provided to discriminate between these diagnoses [13].

Our observations do not weaken the hypothesis that CD4+ lymphocytes in pulmonary sarcoidosis originate from peripheral blood rather than from resident mucosal lymphoid tissue. Indeed, all studies showed that there is a lower percentage of CD4+ lymphocytes presenting the CD103 integrin in BALF of patients with sarcoidosis as compared to patients with other diseases causing pulmonary lymphocytosis. However, as demonstrated in the present study, the BALF CD103+CD4+/CD4+ ratio does not have enough discriminative power to accurately differentiate between sarcoidosis and other causes of pulmonary lymphocytosis, and cannot replace lung biopsy or lymph node needle aspiration for the diagnosis of pulmonary sarcoidosis. One can hypothesize that CD103 integrin expression may be modified under the effect of lung microenvironment or regulatory T cells.

In exploratory analyses, we found that the BALF CD103+CD4+/CD4+ ratio was significantly lower in initial radiographic stage 1 sarcoidosis (median 0.06) as compared to stage 2 (median 0.17). We also found a significant negative correlation between BALF CD103+CD4+/CD4+ ratio and FVC, TLC and DLCO at last visit. Taken together, these data suggest that the BALF CD103+CD4+/CD4+ ratio may be of interest as a prognostic marker in sarcoidosis. This issue should be addressed in future studies.

The authors acknowledge Nastasia Negro for performing part of the data capture and for preparation of figure 1, and Nathalie Bacco for secretarial assistance.