Even though a positive correlation between cold ischemia time and PAGF development has been suggested [3,6-8], other studies found that duration of cold ischemia did not predict outcome after lung transplantation and suggested that other factors interplay with LIRI in PAGF development [9-14]. The question whether LIRI is an independent risk factor for the development of PAGF seems difficult to answer. In clinical studies, often multiple interfering factors are examined simultaneously. Furthermore, a long-term experimental LIRI model, in which PAGF changes can be studied, is missing. The majority of experimental studies use ex vivo LIRI models, like the Langendorff system, which is a non-physiological model and in which it is impossible to investigate reperfusion times beyond the first hours. In addition, an experimental lung transplantation model with the induction of cold ischemia is technically difficult in rodents. Thus, the purpose of this study was to establish an in vivo model of unilateral severe LIRI and to determine whether symptoms resembling PAGF after clinical lung transplantation could be induced. Although the use of warm rather than cold ischemia seems controversial, it has been demonstrated that there are no major differences between short periods of warm and longer periods of cold ischemia [15]. Moreover, warm ischemia has been used extensively in IRI models of liver and kidney as an accelerated model of clinically relevant cold IRI [16-19].

Since most studies have only investigated the early hours of reperfusion [19-32], the effect of severe LIRI up to months after reperfusion is unknown. Furthermore a detailed description of the subset of leukocytes and the time course of infiltration on both short and long term after LIRI is currently missing. Therefore, we have investigated a broad spectrum of LIRI parameters, including lung function, capillary permeability, matrix metallo proteinase (MMP) production, surfactant conversion, and histological changes on the short (days) and long-term (months) after LIRI and we have described leukocyte kinetics.

LIRI has been suggested to be a major risk factor for PAGF. The clinical course of PAGF symptomatically resembles the acute respiratory distress syndrome (ARDS) and can be characterized by different stages, each with their specific clinical, histological and immunological changes [37]. The acute, exsudative phase is featured by a sudden onset of hypoxemia, decreased lung compliance, increased pulmonary artery pressure, and development of non-cardiogenic pulmonary edema [37,38]. Experimental studies have shown that abnormalities in, and depletion of pulmonary surfactant contribute to these symptoms of LIRI [26-28,39-41]. Histological analysis of LIRI shows diffuse alveolar damage with atelectasis, inflammation, intra-alveolar hemorrhage, formation of hyaline membranes and protein-rich edema [1,37]. Finally, production of matrix-metalloproteinases (MMP) is thought to be important in the acute phase of PAGF since MMPs increase the microvascular permeability and thereby enable extravasation of inflammatory cells [42-45].

We found hypoxemia, impaired left lung compliance, a mortality rate of 25% due to development of severe pulmonary edema, and an increase in SA subtype surfactant as early effects of LIRI. Conversion of highly surface active LA into poor surface active SA occurs shortly after reperfusion and is partly due to increased capillary permeability, resulting in influx of serum proteins into the alveolus, as confirmed in our study [28,46]. Serum proteins inhibit surfactant in a dose-dependent manner by competing with surfactant components at the alveolo-capillary barrier [47]. HE slides confirmed extensive alveolar and septal edema, intra-alveolar bleeding, atelectasis and inflammation, which are all indicative of diffuse alveolar damage. The increase in alveolar proteins and neutrophils on day 1 occurred simultaneously with an increased MMP activity. MMP-2 is usually constitutively expressed by endothelial cells, vascular smooth muscle cells and fibroblasts, fitting with the observation in our study of high levels of both pro-, and active MMP-2 found in unoperated animals [42-45,48]. Yano et al demonstrated that LIRI resulted in MMP-9 induction, but not MMP-2 expression 24 hours after LIRI [45]. A higher concentration of MMP-9 was also found in lung edema fluid of ARDS patients [49]. Nevertheless, we found that levels of both pro- and active MMP-9 and MMP-2 per microliter BALf are elevated 24 hours after LIRI, which correlates well with the presence of neutrophils in left lung BALf. Thus 120 minutes of warm ischemia resulted in our experimental study in a mortality rate of 25%, hypoxemia, early impaired left lung compliance, surfactant conversion, diffuse alveolar damage on HE slides and MMP production, which are all features of the exsudative phase of PAGF.

Human lung transplant patients surviving the acute phase of LIRI may either recover from injury or enter a 'chronic' fibroproliferative state, which develops within 4–7 days after the onset of symptoms [37,38]. The progression from an exsudative phase to a 'chronic' fibroproliferative state within one week after LIRI is supported in our study by the presence of fibroproliferative changes on HE slides in the first week after LIRI and an increased number of macrophages, which are important mediators in the regulation of fibroblast function. Importantly, LIRI induced progressive changes resulting in extensive pulmonary injury up to 3 months after reperfusion. This is demonstrated by a decreased number of lymphocytes found in lung tissue on day 30 and 90, impaired left lung compliance up to day 90, extensive atelectasis on HE slides, and a decreased surfactant recycling and secreting capacity of alveolar type II cells reflected by the decreased LA surfactant subtype [50]. Although extensive left pulmonary injury was found on the long-term, hypoxemia was demonstrated up to day 7. Thereafter, no differences in PaO₂ were measured between LIRI animals and controls. This discrepancy may be explained by the fact that PaO₂ was dependent on both lungs, so that the loss of left lung function was compensated by the right lung. Furthermore, even though MMPs are important mediators of pulmonary remodeling, no changes in activity on the long-term were found. It is questionable however whether MMPs are present in the BALf of severe atelectatic and fibrotic lungs. Therefore, in future studies, measurement of MMP activity should also be performed in homogenized lung tissue.

Thus, 120 minutes of warm ischemia in this model induces injury comparable to PAGF and ARDS in clinical lung transplantation on the short, but also on the long-term. Nevertheless, this experimental model has several shortcomings we wish to address. First of all, we used warm ischemia to induce LIRI, whereas in the clinical setting cold ischemic time is associated with PAGF. However, it has been demonstrated that there are no major differences between short periods of warm and longer periods of cold ischemia and warm ischemia has been used extensively in IRI models of liver and kidney as an accelerated model of clinically relevant cold IRI [15-19]. Another disadvantage is that a rather long period of 120 minutes warm ischemia has been used. Shorter periods of warm ischemia have been investigated in a pilot study to setup our model (data not shown) and we found that 120 minutes of warm ischemia is necessary in our hands to induce symptoms comparable to PAGF. This finding is supported by a clinical study by Thabut et al, which shows that the relationship between cold graft ischemic time and survival appears to be of cubic form with a cutoff value of 330 minutes [3]. Thereafter short-term mortality increases rapidly mainly due to development of PAGF [3].

Another goal of this study was to describe leukocyte kinetics following LIRI. The immunologic effects of LIRI have only been studied up to hours after reperfusion, whereas we investigated leukocyte kinetics after LIRI up to 90 days post-reperfusion. Several studies have shown that macrophages are activated during ischemia, followed by the recruitment of neutrophils within hours after the start of reperfusion [16,19,20,22,24,25,29,51,52]. We now add that neutrophil infiltration lasted for 3 days after reperfusion, thereby strengthening the theory that neutrophils are important in perpetuating LIRI. The extended presence of neutrophils after LIRI may be also explained by the fact that phagocytes are important elements of the repair process after LIRI by clearing apoptotic cells and necrotic debris [29]. Nevertheless, since ischemia-reperfusion injury still develops in neutropenic models, it is questionable whether neutrophils are pivotal in LIRI [21,32].

In this regard, other studies suggest an early role for T cells as important mediators of ischemia induced injury. An infiltration of CD4⁺-T-cells occurred in these studies from 1 until 12 hours after reperfusion and disappeared hereafter [16,19,20,51], whereas we found elevated numbers of mainly CD5⁺CD4⁺ and CD5⁺CD8⁺ T-lymphocytes 3 and 7 days after reperfusion in the left lung and TLN. Infiltration and activation of T cells has been classically attributed to the presence of antigen; our findings in an autologous setting may be explained by different mechanisms. First, LIRI induced upregulation of adhesion molecules may attract T cells, such as effector and memory T-cells, which are able to proliferate in an antigen independent fashion by cytokines produced locally (bystander effect), in contrast with naïve T-cells which require antigen presentation by antigen presenting cells [53]. Moreover, antigen specific T-cells may be attracted by released self-antigens [54,55]. Myosin, heat shock proteins and type V collagen, which is released after LIRI in BALf comparable to the level observed in allografted lungs, have been shown to be capable of inducing a T-helper-cell reaction within days after reperfusion [54,55]. The latter is supported by the finding of Waddell and colleagues, who demonstrated upregulation of major histocompatibility II complex after LIRI [56]. Finally, the elevated levels of CD5⁺CD4⁺ and CD5⁺CD8⁺ may be explained by their possible role in the pathogenesis of lung fibrosis, which is also supported by the presence of macrophages in the BALf and lung parenchyma of ischemic animals 3, 7, and 90 days after reperfusion, since they are also thought to be important mediators in the regulation of fibroblast function [37,38,57,58].

Our study furthermore demonstrates an immunosuppressive effect of operation, as measured by the decreased number of lymphocytes in lung tissue on day 1. Although it is very well known that major surgery may cause a short-lasting decrease in blood circulating lymphocytes [59], we now additionally report that thoracotomy causes a one day decrease in the number of lymphocyte subset in lung parenchyma, while the number of lymphocytes in the BALf of sham-operated animals is close to normal. The immunosuppressive effect of surgery may be due to reduced T-cell proliferation and reduced secretion of interleukin-2 (IL-2), IL-4, and gamma interferon by T-lymphocytes, which may be the effect of inhibitory factors secreted by mononuclear phagocytic cells as a result of injury [60]. Moreover, altered migration of memory and activated effector T cells to injury sites may have also contributed to the decreased level of measured cells [61].