Immunodetection of occult eosinophils in lung tissue biopsies may help predict survival in acute lung injury

The patient studies presented in this manuscript were performed in accordance with NIH guidelines and Mayo Foundation institutional policies (Institutional Review Board, 08-001908 Immunohistochemical Study of Eosinophil Degradation in Archived Lung Biopsies), and in compliance with HIPPA guidelines for patient privacy.

An overview of the demographic data and pathological/clinical assessments of our ALI study subjects is presented in Tables 1 and 2. These pulmonary patients were initially identified by study personnel from a search of the Mayo Clinic Arizona Pathology Database with the search key words: lung, biopsy, acute lung injury (ALI), diffuse alveolar damage (DAD), organizing pneumonia, and ARDS. Study personnel subsequently reviewed the available pathology reports and clinical medical records to identify the subset of patients who had received a diagnosis of ALI. Thus, to be included in this study a given patient had to have a diagnosis of ALI and have undergone a biopsy during their course of treatment. It is noteworthy, that this process was not discriminatory and all patients with available biopsy material that had received an ALI diagnosis as part of their standard-of-care were included in this study. The ALI patients included in this study met AECC established clinical criteria [4] for an acute lung injury diagnosis and also displayed characteristic pathological changes linked with this disease. Specifically, assessments of all study patients upon admission with acute respiratory distress revealed the presence of diffuse radiographic abnormalities and hypoxia characterized by an A-a gradient (PaO2/FiO2) less than 300 [4, 32]. Indeed, the PaO2/FiO2 ratio in a subset of these cases (11 of 20 total cases) was below 150 and thus would meet clinical criteria for ARDS [4, 32]. In addition, the twenty ALI patients in this study included thirteen patients (independent of PaO2/FiO2 levels) that required mechanical ventilation during their hospitalization. Review of patient medical records also revealed the absence of typical risk factors associated with ALI, including myocardial infarction, pulmonary embolism, infection/sepsis (e.g., pneumonia), and acute drug reaction. The pathology evaluations of all study-subjects confirmed the clinical indications of ALI. That is, each patient included in our study displayed three specific histopathologies in the available biopsies [33]: (i)The presence of fibrin in the alveoli; (ii)The demonstration of an organizing pneumonia (i.e., a prominent airway cellular infiltrate); and (iii)Evidence of reactive airway epithelial Type II cell hyperplasia. Among these twenty ALI patients, 45% (i.e., 9/20) displayed classical diffuse alveolar damage (ALI-DAD). Seven of the study group did not survive hospitalization, and of these non-surviving patients >71% (5/7) received an ALI-DAD diagnosis. A variety of co-morbid medical disorders were evident in our ALI patients, including several patients with connective tissue disorders. However, examination of the medical records and care-giver notes failed to identify elements of commonality regarding disease onset or progression. Nonetheless, the unresolved character of disease progression in these subjects was such that nineteen of the twenty patients received corticosteroid therapy during their course of treatment.

Control lung samples consisted of uninvolved areas of lung tissue recovered from patients undergoing resection for a diagnosis of lung cancer (Table 1). None of the control subjects were receiving systemic corticosteroids, although four control subjects were receiving inhaled corticosteroids.

The twenty lung specimens obtained from patients with ALI included sixteen surgical lung biopsies and four transbronchial lung biopsies considered adequate for histologic review (i.e., containing alveolated lung parenchyma). All control biopsies were taken from uninvolved areas of surgically resected lung tissue removed for treatment of lung cancer. Lung tissue biopsies were fixed in 4% buffered formalin, embedded in paraffin, and serial 4 μm thick sections were cut.

H&E staining was performed using an automated stain processing unit in the Mayo Clinic Arizona clinical histology unit. Immunohistochemistry was performed using a EPX-mAbas previously described [34]. Evaluations of the slides were performed with either an Olympus BX50 or a Zeiss Axiophot compound microscope.

The eosinophil infiltration of lung tissue using H&E staining was performed in an investigator-blinded fashion independently by an experienced pulmonary pathologist with a specialty in lung diseases (KL) and a pathology resident (PG). The slides were evaluated by each individual as a numerical average calculated from 10 randomly selected hpf (high powered fields; 40x objective/10x ocular lens, 0.29 mm² field of view); that is, a total area of ~3 mm² per biopsy-investigator. The reported values are the mean ± SEM of all investigator-derived counts.

Quantification of eosinophil tissue infiltration within each biopsy using EPX-mAbimmunohistochemistry was independently performed in an intra/inter-blinded fashion that included investigators from pathology (KL, EJ, and PG - EPX-mAbbased eosinophil counts in lung tissue); KL and EJ - EPX-mAbassessments of eosinophil degranulation), a hospital/clinic-based pulmonary fellow (KP - EPX-mAbbased eosinophil counts in lung tissue), and a PhD graduate research fellow (LW - EPX-mAbbased eosinophil counts and degranulation in lung tissue). All evaluations were done at a magnification of 400x. The number of positively stained eosinophils was determined in the alveolar lung parenchyma as a numerical average calculated from 10 randomly selected hpf; that is, a total area of ~3 mm² per biopsy-investigator. The reported values are the mean ± SEM of investigator-derived counts.

The level and extent of eosinophil degranulation observed within each patient biopsy was also determined by scanning 10 randomly selected hpf (40x objective/10x ocular lens, 0.29 mm² field of view); that is, a total area of ~3 mm² per biopsy-investigator. Each field examined was graded using a scale that permitted stratifying the available patients based on a relatively low resolution grading scale that was easily reproduced by multiple evaluators of varying levels of experience and expertise: Level 0 = No identifiable eosinophils and/or degranulation [35]; Level 1a = The field shows evidence of eosinophil degranulation (i.e., extracellular release of EPX) that represents ≤10% of the field's total area and has <3 independent areas within the field displaying degranulation; Level 1b = The field shows similar evidence of eosinophil degranulation as Level 1a but instead displays ≥3 independent areas within the field with evidence of degranulation; Level 2a = The field shows evidence of eosinophil degranulation that includes extracellular release of EPX, enucleated eosinophils (i.e., cytoplasmic fragments), and/or the presence of free eosinophil granules. The extent of degranulation represents 10 - 50% of the field's total area; Level 2b = The field shows similar evidence of eosinophil degranulation as Level 2a but instead has a level of degranulation representing >50% of the field's total area. Eosinophil degranulation was quantified for each of 10 randomly selected hpf of a given biopsy (i.e., patient) by initially applying an increasing numerical value to the level of degranulation evident in the field (Level 0 = 0, Level 1a = 1, Level 1b = 2, Level 2a = 3, and Level 2b = 4). The extent of eosinophil degranulation in the biopsy was then determined as the average of the numerical values assigned to each of the 10 hpf examined. This grading of eosinophil degranulation was performed independently by three outcome-blinded evaluators (an experienced pulmonary pathologist (KL), a pathology resident-fellow (EM), and a Ph.D. graduate student (LW)), all of whom were also unaware of the scores reported by the other evaluators. Degranulation scores are reported as the mean numerical value derived from all three evaluators ± SEM.

Data are expressed as the mean ± SEM. Statistical analysis for comparisons between groups was performed using either a Student's T test or a Wilcoxon Two-Sample Test for non-parametric data for comparisons between data sets that were not uniformly distributed. Differences between mean values were considered significant when p < 0.01. Intraclass correlation coefficients (ICC) were also determined between investigators reading slides as a measure of inter-rater agreement [36].

Serial biopsy sections from either control or ALI subjects were stained for evaluations of eosinophil tissue infiltration. Representative photomicrographs of H&Estained slides as well as lung sections subjected to EPX-mAbimmunohistochemistry are shown in Figure 1. The quantitative evaluations of eosinophil density using both staining methods are presented as individual patient assessments in the histograms of Figure 2. As expected, the evaluation of the H&Estained patient biopsies revealed little evidence of eosinophil infiltration in both control and ALI subjects (0.02 infiltrating eosinophils/hpf and 0.04 infiltrating eosinophils/hpf, respectively). However, evaluation by the same pathology investigators of the serial slides subjected to EPX-mAbimmunohistochemistry demonstrated an enhanced sensitivity to detect eosinophils in these lung tissue sections. Specifically, evaluations of the lungs of control subjects using EPX-mAbimmunohistochemistry revealed a >40-fold increase in the ability to detect tissue infiltrating eosinophils relative to H&Estaining (0.81 eosinophils/hpf vs. 0.02 eosinophils/hpf, p < 0.01).

Evaluation of serial lung sections following EPX-mAbimmunohistochemistry demonstrated that the density of pulmonary eosinophils in the collective group of ALI patients is significantly higher relative to control subjects (3.6-fold, 2.88 eosinophils/hpf vs. 0.81 eosinophils/hpf (p < 0.01), respectively). More importantly, further evaluations of the ALI patients (Figure 2, shaded histograms) surprisingly showed that EPX-mAbdetection of infiltrating lung eosinophils divided these patients into subjects which survived vs. those that did not survive hospitalization (8.4 ±2.9 eosinophils/hpf vs. 1.9 ± 0.6 eosinophils/hpf, p < 0.01).

Assessments of lung sections following EPX-mAbimmunohistochemistry revealed that ALI patients displayed significant and varying levels of degranulation that were quantifiable. This degranulation was often observed in these patients in the absence of identifiable intact eosinophils. The photomicrographs of Figure 3 are representative of the stratified levels of increasing degranulation observed in ALI patients from no evidence of degranulation (Level 0) in a given high powered field to >50% of the field evidencing eosinophil degranulation (Level 2b). Similar to the higher levels of eosinophil infiltration observed in the collective group of ALI patients, the collective group also evidenced a >2-fold increase in the level of eosinophil degranulation compared to control subjects (2.20 ± 0.15/hpf vs. 1.02 ± 0.38/hpf, respectively). More importantly, quantitative assessments of degranulation (i.e., mean numerical score ±SEM) based on EPX-mAbimmunohistochemistry (Table 3 and Figure 4) revealed that ALI patients surviving their hospitalization also displayed significantly higher levels of degranulation compared to non-surviving patients (2.62 ± 0.18/hpf vs. 1.58 ± 0.10/hpf, respectively).