Published ahead of print on February 13, 2003, doi:10.1164/rccm.200205-410OC
© 2003 American Thoracic Society Statin Use Is Associated with Improved Function and Survival of Lung AllograftsDepartment of Medicine and Department of Surgery, University of Pittsburgh, Pittsburgh, Pennsylvania Correspondence and requests for reprints should be addressed to Steven R. Duncan, M.D., Pulmonary, Allergy, and Critical Care Medicine, University of Pittsburgh, 628 NW MUH, 3459 Fifth Ave., Pittsburgh, PA 15213. E-mail: duncsr{at}msx.upmc.edu
3-Hydroxy-3-methylglutaryl coenzyme A reductase inhibitors (statins) are widely used antilipidemic agents that are also immunomodulatory. We evaluated possible effects of these agents after lung transplantation by comparing outcomes of 39 allograft recipients, who were prescribed statins for hyperlipidemia, with those of 161 contemporaneous control recipients who did not receive these drugs. Acute rejection (  Grade II) was less frequently found in the statin group (15.1 versus 25.6% of biopsies, p < 0.01). None of 15 recipients started on statins during postoperative Year 1 developed obliterative bronchiolitis, whereas the cumulative incidence of this complication among control subjects was 37% (p < 0.01). Total cellularity, as well as proportions of inflammatory neutrophils and lymphocytes, were significantly lower in bronchoalveolar lavages of statin recipients. Among double lung recipients, those taking statins had significantly better spirometry: FVC (80 ± 2 versus 70 ± 1%) and FEV1 (87 ± 2 versus 70 ± 1%), as percentages of predicted values, and absolute FEV1/FVC (83.4 ± 1.2 versus 78.6 ± 0.5) (all p < 0.01). The 6-year survival of recipients taking statins (91%) was much greater than that of control subjects (54%) (p < 0.01). These data suggest statin use may have substantial clinical benefits after pulmonary transplantation.
Key Words: bronchiolitis obliterans • graft rejection • hydroxymethylglutaryl CoA reductase inhibitors • transplantation The 3-hydroxy-3-methylglutaryl coenzyme A reductase inhibitors (e.g., statins) appear to have a variety of immunomodulating and other cellular effects, independent of their blood cholesterol lowering, and therapeutic benefits have been suggested for diverse diseases (1–6). The potential utility of statins after lung transplantation, however, is unknown. Prior reports describe improved outcomes among cardiac transplant recipients treated with statins (1, 2). Nonetheless, trials in renal transplantation recipients have not demonstrated important benefits, aside from lowering blood lipid levels (7–9). Moreover, the mechanism(s) that explain drug effects in the cardiac transplantation population have not been fully ascertained. Finally, statin use is occasionally associated with deleterious side effects, and the potential for toxicities may be considerable among lung transplant recipients (10, 11). To evaluate the potential benefits and risks of statins after lung transplantation, we compared at our institution the clinical courses of recipients who were prescribed statins for lipid-lowering effects (LT-statin) with those of a large contemporaneous cohort of recipients who did not receive these agents (LT-control). We found that the LT-statin group had significantly decreased incidences of mild-to-severe acute allograft rejection and obliterative bronchiolitis (OB), along with improved pulmonary function, compared with the control subjects. Overall, maintenance calcineurin inhibitor levels were lower in the LT-statin group, and these recipients also received fewer courses of augmented immunosuppression. Most importantly, the survival of LT-statin group members was much greater than that of untreated recipients. Conversely, toxicities attributable to statins were largely limited to mild and reversible symptoms. These data suggest that statins may have favorable effects in lung transplant recipients.
The study group consisted of 200 consecutive patients who survived more than 30 days after pulmonary transplantation between January 1995 and December 2000. To avoid introductions of biases that could inadvertently occur by conscious (or subliminal) differences in surveillance or treatments of these groups, the observation period ended July 2001, at the time this project was conceptualized and initiated. Pretransplantation evaluations routinely included coronary angiography for males more than 50 years old and for females more than 55 years old. Statins were prescribed to lung transplant recipients during this period solely for treatment of serum total cholesterol exceeding 240 mg/dl, refractory to dietary modification. The choice of statin was at the discretion of the patient's attending physician (Table 1) . Information used for these analyses was abstracted from prospectively recorded databases, under the auspices of Institutional Review Board guidelines at the University of Pittsburgh (Pittsburgh, PA).
Details of patient assessments have been delineated previously (12). In brief, recipients undergo routine scheduled evaluations, including clinical evaluations, blood chemistries, and hematologic indices, spirometry, and fiberoptic bronchoscopy with standardized bronchoalveolar lavage (BAL) (e.g., four aliquots of 50 ml each), and transbronchial biopsies. BAL cell counts are the total count from the entire lavage return. These evaluations were performed at 3-month intervals for the first year, and then every 4 to 6 months thereafter. Additional bronchoscopy, spirometry, or other diagnostic procedures were performed as clinically indicated. Decisions regarding performance of diagnostic procedures were made without consideration of statin use. Acute cellular rejection (ACR) and OB were established from lung biopsies using standard histologic criteria (13). Unless otherwise stated, spirometric parameters were referenced to the predicted values of the recipients. Spirometric evaluation for bronchiolitis obliterans syndrome (BOS) was based on changes in FEV1, as described elsewhere (14). Cytomegalovirus (CMV) pneumonitis was defined as identification of characteristic viral histologic findings and positive viral cultures (12). A standardized immunosuppressant regimen was employed during the study period. Recipients were initially treated with tacrolimus, with dose titrations to maintain whole blood levels of 15–20 ng/ml. Tacrolimus-intolerant patients were prescribed a cyclosporin A (CsA) formulation titrated to serum levels of 250–300 ng/ml. Calcineurin inhibitor (CsA or tacrolimus) doses were routinely evaluated by frequent blood level determinations, with adjustments influenced by clinical events (i.e., absence or presence of rejection). In general, calcineurin blood levels at the lower limits of therapeutic were tolerated in recipients without significant recent rejection, whereas higher normal or slightly supratherapeutic levels were targeted for those with recent or severe rejection. Unless contraindicated by leukopenia, azathioprine was administered at 2 mg/kg. Patients with refractory leukopenia or, rarely, refractory rejection were switched to mycophenolate at a dose of 250 mg twice daily and increased to 1,500 mg twice daily as tolerated by leukopenia. Prednisone was initially prescribed at 20 mg/day and gradually tapered, in the absence of interval rejection episodes, to 10 mg/day or less within the first postoperative year. Episodes of moderate-to-severe acute rejection or active OB were initially treated with pulses of methylprednisolone (1 g/day) for 3 days, along with increases in daily prednisone maintenance. Rejection episodes that were refractory to corticosteroids were treated with antilymphocyte globulins (OKT-3 or rabbit antithymocyte globulin). Treatment decisions, including immunosuppressant dosing, were made without consideration of statin use. For purposes of this investigation, and unless otherwise specified, all recipients who received statins at any time after transplantation were stratified into the LT-statin population. Thus, parameters of this population were typically analyzed irrespective of whether individual LT-statin subjects had actually yet received or remained on drug at the time of data accrual (analogous to intent-to-treat).
Two group comparisons of continuous and parametrically distributed values were made by unpaired t test, with Bonferroni correction. Analogous comparisons using nonparametrically distributed values were made by Mann–Whitney test. Dichotomous variables were evaluated by
Subjects Characteristics of the study populations are delineated in Table 2 . LT-statin subjects were slightly older, more likely to have had emphysema, and less likely to have had cystic fibrosis, compared with control subjects. The median time for starting statins was 1 year after transplantation (mean, 1.6 ± 0.2 years; range, 0–5.0 years). The mean duration of statin therapy was 2.0 ± 0.2 years. LT-statin subjects were evenly distributed throughout the study period, as indicated by comparisons of transplant sequence number (the unique, sequential number assigned to successive lung transplantations at this center) (i.e., 507 ± 6 versus 504 ± 12, LT-control and LT-statin, respectively). The mean observation duration for the study subjects was 3.0 ± 1.9 years (range, 0.1 to 6.4 years). The duration of observation after transplantation was significantly greater among LT-statin subjects (3.6 ± 0.3 years) than among control subjects (2.8 ± 0.2 years, p < 0.02), a possible result of intergroup survival differences (see below).
Allograft Rejection The overall incidence of mild or greater (Grade 2 or more) ACR diagnosed by lung biopsy (12) was significantly higher among LT-control than among statin-treated recipients (26 versus 15% of lung biopsies, p < 0.01) (Figure 1) . This intergroup difference was particularly striking when analyses were limited to more severe ACR Grades 3 and 4 (13 versus 4% of biopsies, LT-control and LT-statin, respectively, p < 0.01).
However, because of the longer duration of observation of LT-statin recipients, in turn a probable function of their improved survival (see below), a greater proportion of lung biopsies in this subpopulation were performed in later periods after transplantation: the mean time of biopsy was 2.0 ± 1.5 years after transplantation for LT-statin recipients versus 1.1 ± 1.2 years among LT-control subjects (p < 0.05). Given the increased frequency of later biopsies among LT-statin recipients, and the generally decreasing incidence of high-grade ACR with the passage of time after transplantation (15), it seemed possible that the apparent differences in ACR incidences could be due to an observational bias. Accordingly, we reexamined occurrences of ACR within a fixed period of time (1 to 4 years after transplantation). The mean timing of lung biopsies was equivalent in both groups during this particular observation period (2.2 ± 0.1 versus 2.0 ± 0.0 years, LT-statin and LT-control, respectively) and there were similar frequencies of biopsies performed (12.7 versus 11.3 biopsies per patient, LT-statin and LT-control, respectively). As expected, the incidences of ACR after the first posttransplantation year were slightly less in both groups. There remained a trend for more frequent ACR of Grade 2 or more in the LT-control group (18%) relative to the LT-statin group (12%), but these differences did not reach statistical significance, a likely function of the smaller number of biopsies performed in later years. Despite this limitation, however, the intergroup incidences of more severe ACR (Grades 3 and 4) was fourfold greater in the LT-control group (8%) than in the LT-statin group (2%), and this difference was significant (p < 0.02).
Chronic Allograft Rejection
Statins were administered to recipients solely to treat hypercholesterolemia, without considerations that these drugs may have other effects. To ensure that the beneficial effects of statins on chronic rejection and other outcome parameters are not somehow attributable to a fortuitous stratification, we performed intergroup comparisons of risk factors that may influence graft survival and posttransplantation outcomes (12, 16–21). As shown in Table 3 , there were no significant differences between LT-statin and LT-control groups with respect to donor–recipient major histocompatibility (MHC) mismatches, or donor organ ischemic time or age. Similarly, donor–recipient CMV serologic status most predictive of serious CMV infections (donor positive, recipient negative) (12) was equivalent in both groups. Serologic status most often associated with CMV infections (either donor or recipient positive) were most frequent in the LT-statin group (Table 3).
Immunosuppression We hypothesized that at least some of the apparent beneficial effect(s) of statin therapy on allograft rejection might also result in measurable decrements of immunosuppression, because these medications are adjusted in accordance with clinical events. If so, moreover, these particular data would be additional corroboration that statin use has favorable actions among lung transplantation recipients. We found that statin administration was associated with decrements in both long-term maintenance immunosuppression doses (Figure 3) , as well reductions in frequencies of boost or augmentation immunosuppressant treatments (Figures 3 and 4) .
BAL Cell Counts We evaluated the cellularity of BAL fluid, as another independent measure of statin effects on lung allograft rejection. Total BAL cell counts were reduced in LT-statin subjects compared with control subjects (16.7 ± 1.3 versus 26.4 ± 1.0 million cells, p < 0.001). Given that intergroup differences in incidences of rejection or infection could bias these results independently of subclinical inflammation (22), we also performed an analysis of BAL cellularity using procedures wherein the recipients were shown by other means (biopsies and negative microbiological stains and cultures) to have no evidence of either rejection exceeding ACR Grade 1 or infection. Total BAL cell counts among LT-control subjects remained significantly greater than among LT-statin recipients (11.8 ± 1.7 versus 17.4 ± 0.8 million cells, p < 0.01). The quantities of BAL fluid returns are not permanently recorded at our institution and, hence, it seems possible that the greater total cellularity among LT-control subjects could result if returned volumes were increased in these patients. However, we believe it far more probable that these data are likely to be highly biased to minimize intergroup differences, because the BAL return tends to diminish in patients with OB or severe expiratory obstruction (more of whom were LT-control subjects). In any event, LT-statin subjects also had lower percentages of inflammatory cells in their BAL compared with control subjects, notably including neutrophils (14.7 ± 1.5 versus 18.8 ± 0.6% of BAL cells, p = 0.03) and lymphocytes (7.2 ± 0.2 versus 8.7 ± 0.7% of BAL cells, p < 0.001) (22).
Pulmonary Function Tests Given the potential for introduction of observational biases due to possible intergroup differences in assessment frequency or duration of follow-up, we conducted other analyses of PFTs with assessments at finite posttransplantation periods. Results of these data compilations and comparisons are depicted in Figure 5 . Whereas PFT parameters for both LT-statin and LT-control subjects were nearly identical 1 year after transplantation, values for FVC, FEV1, and FEV1/FVC were all significantly greater among the former at subsequent time points. Intergroup values at the latest time point in this analysis (Year 4) tended to converge, but numbers of observations were comparatively limited (these data reflect 67 individual PFT evaluations in this population); however, significant differences remained evident even at the late time point.
To further assess the effects of statins on lung function, we also performed analyses of spirometric values limited per se to the LT-statin population, to eliminate potential cryptic confounding due to differences in underlying diseases or unrecognized differential treatments. The control group in these analyses consisted of LT-statin recipients who were not actually receiving the drug at the time of the individual PFT. In most cases, statins had not yet been prescribed, but in a small number (n = 4) these drugs had been discontinued for various reasons (see below). The comparison experimental group here was composed of recipients receiving statins at the time of measurement. As depicted in Figure 6 , the expected rate of PFT decline over time was significantly less when these patients were actually taking statins. This difference was especially striking among DL and HL recipients, but the rate of PFT decline while taking statins was also less among the widely disparate (e.g., increased noise-to-signal ratio) single lung transplant (SL) recipients. We limited the analyses interval here to Years 1–4, to minimize variations in the PFT times. The mean time of PFT testing during these intervals for DL and HL LT-statin subjects was somewhat later for patients receiving the drugs than for those not receiving them at the time of testing (2.4 ± 0.1 versus 2.0 ± 0.1, p = 0.03). Because spirometric values tend to deteriorate with time after transplantation (and OB prevalence increases), the differences in observation timing here should bias against the LT-statin recipients who were actually receiving drugs. There was no difference in the mean time of PFT testing for single lung recipients while receiving or not receiving statins (2.2 ± 0.1 versus 2.1 ± 0.1, p = NS).
We also analyzed the freedom from spirometrically determined BOS. Again, because of the extreme diversity of native, contralateral lung physiology among SL recipients, the signal-to-noise ratio for PFT data from DL and HL recipients was less extreme, and thus more conducive to statistical comparisons. Two of nine DL and HL LT-statin subjects exhibited PLT decrements consistent with mild BOS (Stage 1). This afflicted proportion (22%) was about one-half that seen among comparable DL and HL LT-control subjects (Figure 7) . Given the small numbers of LT-statin subjects, however, these differences did not attain statistical significance. None of the LT-statin DL and HL recipients developed more severe BOS 2 or BOS 3, resulting in a statistically significant intergroup difference compared with LT-control subjects (Figure 7).
Survival The 6-year survival of all recipients who took statins at any point after transplantation (91%) was significantly better than that of LT-control subjects (54%, p = 0.002) (Figure 8A) . This survival difference is not explained by intergroup differences of risk factors related to donor–recipient mismatches or allograft procurements (Table 3). Given that this intergroup survival difference was striking, we performed a number of other analyses to ensure these findings were not merely due to confounding biases or unrecognized epiphenomena.
Because statin therapy tended to be started several months after transplantation, we were concerned that evaluating a group of patients who were well along in their course at the time of drug prescription (LT-statin) could introduce a bias. In effect, we could be comparing a possibly more mature subpopulation (e.g., the LT-statin group), which had already survived several months (and thus may be selected for subsequent survival), with a distinctly different pool of recipients that included those who would not live long enough after transplantation to develop hyperlipidemia (and thus could not be assigned to the LT-statin pool). Accordingly, we reanalyzed survival after limiting the study populations to LT-statin and LT-control subjects who had already survived 2 years. Subsequent survival was better in both of these subpopulations compared with the former analyses that included all recipients (as predicted) but, again, a prescription for statins apparently conferred a significant survival advantage (Figure 8B). Of considerable note, only 1 of these 30 LT-statin recipients succumbed during the subsequent observation period. Other analyses seem also to exclude the possibility that intergroup demographic differences including age, diagnosis, type of transplant, or calcineurin regimen could impart enough bias to confound these survival data. As an example, to exclude differences in survival being attributable to the slightly greater age of LT-statin recipients, we separately analyzed survival for groups of patients either younger or older than 49 years (the mean age of the LT-control group). However, age did not have any seeming correlation with survival in our cohorts (p = 0.74). Similarly, there were disparate proportions of patients with emphysema and cystic fibrosis in our two groups, but we were unable to find any independent survival difference between these two cohorts (p = 0.55). The type of transplantation could conceivably bias survival, and there was a greater proportion of single lung transplant recipients in the LT-statin group. Nonetheless, the only identifiable trend (albeit not statistically significant) for greater survival was seen among DL recipients (67 versus 59% 6-year survival for SL recipients, p = 0.15) and this correlation, if real, would negatively bias results for the LT-statin group. Similarly, we found no apparent difference in survival that could be attributable to the selection of calcineurin inhibitors (p = 0.64). Although not necessarily a confounding bias, the survival advantage for LT-statin subjects could conceivably be attributable to a direct antilipidemic action or another reduction in mortality due to ischemic cardiac disease. However, these patients were intensively screened for coronary artery disease before transplantation. Moreover, we are aware of only one cardiac ischemic event that occurred among the study patients, and this was a nonfatal subendocardial myocardial infarction in an LT-statin recipient. A review of the causes of mortality did not reveal any cardiac events (Table 4) , but there were seven undiagnosed deaths in the control group (these occurred in other, geographically remote institutions). Reanalyses after deletion of these particular seven LT-control subjects improve survival in this subpopulation to 58% at 6 years, but this is still significantly less than the 91% survival among the LT-statin recipients (p < 0.01).
The causes of excess relative mortality in the control group are largely attributable to complications of allograft rejection and/or the immunosuppression regimens used to prevent or treat rejection (Table 4). The majority of deaths in this cohort were immediately preceded by or concomitant with either moderately severe and/or recurrent ACR (n = 13), or progressive or severe OB (n = 14), that were believed by the recipients' attending physicians to have been the cause or a significant contributor to the mortality. Infections and neoplasms (including lymphomas) were the most numerous proximate cause of death, and these complications were much more frequent among the LT-control subjects. The frequency of pulmonary infections in general, aside from the lethal episodes delineated in Table 4, was also much less among the statin-treated recipients. Only 3.2% of LT-statin lung biopsies showed histopathologic evidence of pneumonia, versus a 7.4% incidence among control subjects (p = 0.03). CMV pneumonitis developed much less often among LT-statin subjects (2.6%) compared with occurrences in 26 (16.1%) of LT-control subjects (p = 0.03).
Toxicity Serum creatinine was slightly elevated in statin recipients (1.9 ± 0.1, n = 222) compared with control subjects (1.7 ± 0.0 mg/dl) based on determinations routinely made at the time of lung biopsies after transplantation (p = 0.001). However, serum creatinine values rise progressively with time after transplantation, and the LT-statin recipients lived longer and thus had a greater proportion of later creatinine determinations. When observation intervals were equalized by limiting analyses to Years 1 to 4 after transplantation, serum creatinine levels were identical among both groups (1.9 ± 0.1 mg/dl).
These data strongly suggest that administration of statins (3-hydroxy-3-methylglutaryl coenzyme A reductase inhibitors) confers substantial clinical benefits in pulmonary allograft recipients. Statin treatment was associated with a lesser frequency of acute allograft rejection, as well as with significant reductions of chronic rejection (OB) and high-grade BOS. BAL returns in LT-statin recipients also had fewer immune effector cells, an independent but corroborative indication that these recipients had reductions of intragraft inflammation (22). As a likely consequence, the pulmonary function of the LT-statin group was also significantly better than that of the control cohort. Furthermore, these favorable effects among the LT-statin recipients were achieved with lower levels of maintenance global immunosuppressives and less frequent treatments with intense antirejection therapies. In turn, these lower levels of nonspecific immunosuppression, perhaps in conjunction with less frequent or less severe allograft injuries, may account for the significant reductions of lethal infections and neoplasms, and much improved overall survival, among the LT-statin recipients. Remarkably diverse pharmacologic effects have been attributed to the statin-induced inhibition of mevalonate synthesis. Mevalonate is an important intermediate metabolite of multiple isoprenoids, including farnesyl- and geranylgeranyl-pyrophosphates. These isoprenoids are a source for the hydrophobic prenyl groups that covalently bind to and activate Ras GTPase proteins (23). In turn, Ras superfamily proteins are critical regulatory factors of signal transduction that have myriad, complex effects on cell differentiation, proliferation, and apoptosis (24). Thus, statins have the potential to alter transcriptional regulation of pleotropic gene products that influence seemingly unrelated cellular mechanisms. In particular, statins have been reported to exert a large number of highly varied antiinflammatory effects. At least some of the statins seem capable of binding to or otherwise blocking interactions between leukocyte function-associated antigen-1 and intercellular adhesion molecule-1 located on endothelial surfaces (25, 26), In addition, basement membrane degradation by lymphocytes is likely suppressed because of inhibition by matrix metalloprotease-9 (27). Functional effects of these actions could result in inhibition of cell adherence and diapedesis across the endothelium. Cognate T cell–proliferative responses to alloantigens, an important component of rejection mechanisms (28), may be directly inhibited by statins, and the drugs also reportedly alter cytokine production (6) and apoptosis thresholds (29). In addition, statins have been shown to be capable of downregulating cytotoxic responses of natural killer cells (1), and the drugs may also have multiple inhibitory effects on critical steps in various acute-phase reactions (30). The relative importance of these individual actions in allograft recipients has not been ascertained. The aforementioned immunomodulatory effects would likely result in relatively nonspecific immune suppression that could favorably act to supplement conventional antirejection medications, and would further suppress allograft rejection responses. Nonetheless, if a global immunosuppressive effect were indeed the primary operative mechanism of statin treatments, it also seems most likely that the LT-statin recipients would also exhibit increased problems with infectious and/or neoplastic complications, although present observations suggest the contrary.
The clinical findings here that the LT-statin recipients have both decreased incidences of allograft rejection and fewer immunosuppression complications are perhaps best reconciled with reports that these drugs inhibit the facultative expression of MHC Class II molecules (31). Most nucleated cells other than dedicated antigen-presenting cells typically have limited MHC Class II expression under usual conditions. However, many of these same tissues, notably including donor endothelial cells and/or airway epithelium, upregulate their expression of MHC Class II in the presence of cytokines, primarily interferon- We were careful to attempt to eliminate methodologic approaches or analyses that could recognizably or conceivably introduce bias. Indeed, we intentionally performed several analyses, especially those concerning recipient survival, to eliminate possible subject selection biases. In other cases we intentionally performed analyses that were biased to minimize treatment effects. Despite these considerations, the data robustly show beneficial effects of statin treatments in lung transplant recipients. Moreover, measures that are only obliquely interdependent (e.g., histopathology, BAL, drug levels, PFTs, and overall survival) are consistent in their demonstrations of statin benefits. Nonetheless, it is worth emphasizing that this was essentially a retrospective study and, as such, is liable, despite our best intentions otherwise, to suffer from inherent limitations and/or unrecognized biases. Thus, the present findings should be construed as indicative, rather than proof, that statins are associated with favorable treatment effects in this transplant population. We believe these data should be further corroborated by prospective randomized trial before advocacy of widespread statin prescription to lung transplant recipients. In summary, these data suggest, by multiple, loosely interdependent measures, that administration of statins to lung transplant recipients results in fewer episodes of and less severe allograft rejection, decreased relative risks of infectious or neoplastic complications, and improved overall survival. These effects are perhaps most consistent with an action of statins that specifically ameliorates responses to allografts, but largely leaves other immune responses largely intact. Although the putative advantages for statin use in lung transplant recipients appear to have considerable potential clinical importance, the present data need confirmation in prospective randomized trial.
Supported in part by US NIH grant 1RO1HL64192. Received in original form May 9, 2002; accepted in final form February 5, 2003
This article has been cited by other articles:
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
METHODS
2 or Fisher exact test. Changes in pulmonary physiologic parameters as functions of time were compared by both analysis of covariance for differences between slopes and by Komolgorov–Smirnov analyses for distribution differences. Survival durations were evaluated by the Kaplan–Meier product-limit method, with analyses by log-rank test, unless a high proportion of censored values necessitated log normal analysis. Significance was defined as p < 0.05. Data are shown as means ± SEM. 
elaborated by T cells (
-I domain. J Mol Biol 1999;292:1–9.
