Transplantation is the mainstay of treatment for patients with end-stage lung disease refractory to medical therapy, yet its median survival remains limited due to the high incidence of primary graft dysfunction (PGD) and allograft rejection (AR).1 Inflammation within the donor, cold ischemic time, and ischemia reperfusion injury (IRI) have all been implicated in the development of PGD and AR.2-4 Upon death, a catecholamine and cytokine storm occurs within the donor that results in an inflammatory cell infiltration of the lungs, with subsequent upregulation of MHC class II expression within the graft.5 Donor lungs then undergo a preservation flush and are stored on ice for transport before implantation within the recipient, as per standard donor organ procurement. This ischemic insult leads to the generation of reactive oxygen species with dysfunctional mitochondrial respiration that, in association with altered cellular metabolism, results in cell injury.3 The donor lung leukocyte compartment is activated, secreting proinflammatory cytokines and chemokines. This is exacerbated upon reperfusion, with further reactive oxygen species production and activation of matrix metalloproteinases and other proteases.6,7 Cell death ensues via necrosis, necroptosis, and apoptosis, with the release of proinflammatory damage associated molecular patterns.7 This cascade of events causes cellular injury within the lungs, manifesting as PGD. This inflamed and immunologically activated environment within the donor lung promotes AR, driving recipient allorecognition through donor leukocyte activation and further upregulation of MHC class II expression on donor tissue.4,8 Understanding the complex pathophysiological signaling processes involved in this donor lung inflammation and injury is essential before the development of preventative or interventional therapies.
Ex vivo lung perfusion (EVLP) was initially developed as a technique to enable the extended evaluation of donor lungs before transplantation;9 however, it has since been used as a reconditioning tool for marginal donors.10 The physiological preservation of donor lungs using EVLP may reduce the impact of IRI and has been shown to be associated with a reduced incidence of PGD posttransplantation.11,12 Interestingly, although the INSPIRE trial demonstrated a reduced incidence of PGD3 within 72 hours of transplant, no short-term survival benefit was observed compared with standard transplantation.13 However, it remains to be seen if this can impact on the long-term outcomes after transplant. We have previously demonstrated that EVLP removes a proportion of donor leukocytes before transplant, altering the immunogenicity of the graft. In this work, we reported that EVLP transplantation resulted in a reduction in T-cell recruitment in the early postoperative period, compared with static cold storage (SCS).14 Anecdotally, EVLP may impart additional benefits via modulating cellular and molecular pathways within the donor lung; however, this has yet to be evaluated. In this study, we profiled a range of inflammatory signaling pathways within the donor lung following standard versus EVLP transplantation.
MATERIALS AND METHODS
The Ethics Committee for Animal Research Malmö/Lund approved the study (No. M174-15). The animals were treated in accordance to the “Principles of Laboratory Animal Care” by the National Society of Medical Research, and the “Guide for the Care and Use of Laboratory Animals” by the National Institutes of Health 1985. The study design and article have been prepared according to NC3Rs guidelines for reporting animal research.15 In keeping with these guidelines, samples collected during our previously reported animal transplant model were utilized to evaluate the impact of EVLP on the inflammatory signaling profile of the donor lung posttransplantation.14
Twenty-four Swedish outdoor free-range domestic pigs (n = 12 male donors versus n = 12 female recipients) with a mean weight of 63 kg ± 2 kg were used in this experiment as previously described.14 All animal studies were performed by the Lund University and Skåne University Hospital team. Collected samples were then stored and transported to the University of Manchester for further analysis.
Donor Organ Procurement
Male donor lungs (n = 12) were retrieved following standard donor organ procurement guidelines and using the surgical technique previously described in porcine transplant models.16 Briefly, the PA and aortic arch were cannulated with distal ligation of the brachiocephalic and left subclavian arteries. The superior and inferior vena cavae were ligated, and the ascending and descending aorta were clamped. A small cut was made in the right atrium and the left atrial (LA) appendage. The PA was then perfused with 2 L of 4°C Perfadex (XVIVO Perfusion AB, Gothenburg, Sweden), and the heart-lung block was excised. The left lung was retrieved at the level of the carina and the orifices of the left pulmonary veins. The lung was then submerged in cold (6–8°C) Perfadex solution (XVIVO Perfusion AB, Göteborg, Sweden) for a period of 2 hours. The lungs were then randomly assigned to either standard (n = 6) or EVLP (n = 6) groups. The standard group were transplanted following 2 hours SCS. In the EVLP group, the lungs underwent 2 hours SCS followed by 3 hours of EVLP, before transplantation (Figure 1). The aim of the experiment was to evaluate if inflammatory and molecular signaling processes are altered following EVLP versus SCS lung transplantation. Therefore, a minimal ischemic time of 2 hours was used, to avoid significant donor lung injury and ensure successful transplantation with stable postoperative lung function in both groups.
Ex Vivo Lung Perfusion
EVLP was performed for 3 hours. The technique followed the previously described Steen protocol,17 using the LS1 system (XVIVO Perfusion AB, Göteborg Sweden). Briefly, following removal from Perfadex, the lung was inspected, and any visible clots in the PA and LA were removed. The PA was then cannulated, and the LA appendage was left open. The trachea was cleared of secretions, and a size 9-mm endotracheal (ET) tube was inserted. The lung was then placed in the organ perfusion chamber, and the PA cannula was connected to the circuit. Following the removal of air, the lungs underwent controlled cellular reperfusion via the PA, with gradual rewarming and protective volume-controlled ventilation. To achieve a cellular perfusate, STEEN solution was supplemented with leukocyte-depleted red blood cells, obtained from separate pigs with the same blood group, to achieve a target hematocrit of 15%. After 3 hours, perfusion flow and ventilation were reduced while the lung underwent core cooling to 20°C. Perfusion and ventilation were then discontinued, and the PA cannula and ET tube were removed, with the lung clamped in a semi-inflated state. The lung remained immersed in the perfusion chamber with topical cooling until transplantation.
Left Single Lung Transplantation
Female recipient pigs (n = 12) were randomized to receive a left single lung transplantation using either standard or EVLP lungs. All transplantations were male to female to determine the presence of donor-derived DNA via detection of the Y chromosome. Left lung transplantation was performed, followed by right pneumonectomy within 1 hour, following the surgical technique previously described in detail by Steen et al.16,17 Following transplantation, animals were turned with their left side up, and ventilation and fluid resuscitation were maintained for 24 hours. At 24 hours, the pigs were sacrificed following a final assessment of pulmonary hemodynamic and gas exchange. At no point in the study was full anesthesia withdrawn.
Sample Collection and Storage
Following transplantation, 20-mL samples of peripheral blood were taken at sequential time points (0, 6, 12, and 24 h). At 24 hours, lung tissue biopsy samples were taken. Following venesection, all peripheral blood samples were centrifuged at 500g for 10 minutes at 4°C, and plasma supernatant aliquots were stored at –80°C. The lung biopsy samples taken at 24 hours posttransplant were snap-frozen in liquid nitrogen and stored at –80°C.
A human phosphokinase proteome profiler (ARY003B) and a human apoptosis antibody array kit (ARY009) were used (R&D Systems, Minneapolis) to examine the relative expression of 35 apoptosis-related proteins and the phosphorylation of 44 kinases in the transplanted lung tissue at 24 hours. Using a BLAST search, homology of over 90% was determined between porcine and human proteins, and therefore, cross-reactivity was predicted in 29 of the proteins.
Proteomic profiling was performed according to manufacturer’s guidelines. Briefly, 100-mg porcine lung tissue collected at 24 hours posttransplantation was weighed out and homogenized in lysis buffer to create lung tissue lysate. The amount of protein present was then quantified using a bicinchoninic acid protein assay (ThermoFisher Scientific, Waltham, United Kingdom), and all samples standardized to the lowest protein concentration. Samples were pooled into 2 sets of 3 samples, using a total of 2 membranes per group (EVLP versus SCS). Variability of expression was low across the membranes within each group (Figures S1 and S2, SDC, http://links.lww.com/TP/B952). The membranes were blocked, and the tissue lysate and relevant detection antibodies were added following the appropriate incubation and wash steps, as per the protocol. A chemireagent mix was then added, and chemilluminescence was used to visualize the membranes using a Bio-Rad ChemiDoc MP imaging system (Bio-Rad Hertfordshire, United Kingdom). Negative controls (PBS) were included within the kit to allow nonspecific binding of proteins to be identified. Pixel density analysis was performed using ImageJ software and normalized using the positive control on each membrane. Differential expression was only accepted when there was >10% change in pixel density between the 2 groups. We selected a low-arbitrary threshold as the study groups were highly defined, that is, age- and weight-matched recipients with identical surgical procedures other than the intervention.
Quantitative polymerase chain reaction (qPCR) was performed to determine the amount of cell-free genomic DNA (gDNA) and mitochondrial DNA (mtDNA) present in the plasma samples collected during the first 24 hours posttransplant. The DNA was extracted from stored plasma samples using a QIAamp DNA Mini and Blood Mini Kit (Qiagen, Manchester, United Kingdom), according to the manufacturers protocol. To identify DNA, primers were designed using the Primer Express Software v3.0.1 (Life Technologies, Paisley, United Kingdom) and their homology to other genes assessed using BLAST. To identify cell-free gDNA, forward and reverse primers to glyceraldehyde-3-phosphate (GAPDH) were used (GAPDH forward 5′ TGCTCCTCCCCGTTCGA 3′, GAPDH reverse 5′ GGCTTTACCTGGCAATGCA 3′). To identify donor DNA via the Y chromosome, primers specific for the sex-determining-Y (SRY) region were used (SRY forward 5′ CAAGTGGCTGGGATGCAAGT 3′, SRY reverse 5′ TCGAAGAATGGGCGCTTTT 3′). To identify cell-free mtDNA, forward and reverse primers specific to cytochrome B were used (cytochrome B forward 5′ ACACATCAGACACAACAACA 3′, cytochrome B reverse 5′ GTAGCGAATAACTCATCCGTAA 3′). All primers were synthesized and desalted by Sigma Aldrich and then resuspended in nuclease-free water (Ambion) and their concentration adjusted to 150 nM.
The extracted plasma DNA was mixed with Power SYBR green PCR master mix (Life Technologies, Paisley, United Kingdom), nuclease-free water and the relevant forward and reverse primers. qPCR was then performed using a QuantStudio12K Flex System. Samples with Ct values of >40 were classified as having undetectable levels of DNA.
We utilized a model-informed experimental design with a defined control group (SCS). Statistical analysis was performed using SPSS v.22.0. Data normality were determined by using the Shapiro-Wilk test, and normally distributed data were expressed as mean ± SD (SD). Two-way ANOVA with post hoc multiple comparisons enabled analysis of changes in cell-free DNA between groups overtime. Independent samples T-test or the Mann-Whitney U test were then utilized for a direct comparison between EVLP and standard samples obtained at a single time point, dependent on data distribution. For the proteomics data, descriptive statistics were utilized to identify any changes in the relative expression/phosphorylation between SCS and EVLP transplanted lungs. GraphPad Prism v.7.0 was utilized to formulate graphs.
All pigs remained hemodynamically stable within the first 24 hours posttransplantation, with a Pao2/FiO2 ratio maintained above 40 kPa throughout (for further details of posttransplant graft function14). At 6 hours, the mean Pao2/FiO2 ratio was marginally lower in the EVLP group (54.3 ± 7.4 kPa) compared with the standard transplant group (55.4 ± 4.6 kPa), although this was not significant. However, at 12 and 24 hours, the mean Pao2/FiO2 ratio was higher in the EVLP group compared with the standard transplant group (12 h: 73.5 ± 1.9 kPa versus 56.1 ± 4.9 kPa; 24 h: 63.1 ± 1.6 kPa versus 50.7 ± 9.4 kPa). This difference was significant at 12 hours (P < 0.01).
The relative level of phosphorylation of 29 kinases were analyzed, and the expression of 19 altered by over 10% between the EVLP and SCS groups (Figure 2). The expression of the remaining kinases fell below the 10% threshold set between the groups (Figure S3, SDC, http://links.lww.com/TP/B952). There was a general upregulation of phosphorylation observed in the EVLP transplanted lung tissue, with a differential expression of over 20% in AMPKalpha1 (43.89%), Hck (35.71%), Fgr (33.73%), PDGFRbeta (29.64%), STAT5a/b (27.39%), Lck (23.92%), STAT6 (22.01%), FAK (21.76%), and ERK1/2 (20.39%). In addition, of the 35 proteins associated with cell death and survival, the expression of 11 altered by over 10% between the EVLP and SCS groups (Figure 3). The expression of the remaining 24 did not change by >10% between the EVLP and SCS groups (Figure S4, SDC, http://links.lww.com/TP/B952). Within the EVLP transplanted tissue, there was a marked downregulation of clusterin (28.98%), cytochrome C (19.54%), and BCLx (14.86%), and to a lesser degree, BAX (11.23%), and HTRA2/OMI (11.05%). In addition, there was a marked upregulation of HSP-70 (23.17%), Procaspase 3 (17.08%), Trail R1/DR4 (15.64%), Trail R2/DR5 (12.62%), and BCL-2 (11.08%).
There was no statistically significant difference in the circulating levels of cell-free gDNA or mtDNA overtime posttransplant between the EVLP and standard lung transplant recipients (Figures 4 and 5; P = 0.1670 and P = 0.3406, respectively). Additionally, no cell-free Y chromosome was detected at any time point. However, there was significantly less circulating cell-free mtDNA in the EVLP group within the early postoperative period (0 h—within minutes of cross-clamp removal and reperfusion) (Figure 5; P = 0.016). After this time point, in the EVLP group, concentrations peaked at 6 hours before declining overtime. Conversely, in the standard group, concentrations steadily increased in the first 12 hours posttransplant before rapidly declining to levels similar to the EVLP group (Figure 5).
Lung transplantation remains limited by the high incidence of morbidity and mortality secondary to rejection and PGD.18 Donor lungs are exposed to a myriad of pathophysiological processes that occur upon donor death and during ischemic storage, with resultant activation of inflammatory cascades.3,19 This contributes to the inevitable IRI response further exacerbating recipient inflammatory infiltration with subsequent graft dysfunction.3 In addition to facilitating the extended evaluation of donor lungs,17 EVLP can also be used to recondition suboptimal donor lungs and potentially reduce graft inflammation before transplantation, ameliorating graft dysfunction posttransplantation.10-12 However, the cellular and molecular mechanisms that underpin this process remain poorly understood.
In this preclinical transplant model, we identified that EVLP resulted in a universal upregulation of protein expression and increased phosphorylation of a range of kinases within the lung tissue following 24 hours of transplantation, suggesting a complex interaction between wide ranging pathways involved in cell signaling. Such universal upregulation was not observed following transplantation with lungs that had only undergone SCS. Furthermore, the expression of proteins involved in apoptosis and necrosis was notably improved following EVLP compared with standard transplantation. There was increased relative phosphorylation of protein kinase B and PRAS40 within the EVLP transplanted lung, both of which regulate processes involved in cellular proliferation and survival and inhibit apoptosis. AKT inhibits the proapoptotic Bcl-2 family proteins, BAX and Bak, and along with PRAS40, also inhibit Bad, thereby impairing the activation of apoptotic initiator proteins.20 In addition, through downstream binding and phosphorylation of FoxO, AKT inhibits the activity of FoxO transcription factors that normally induce the expression of the proapoptotic Bcl-2 family proteins and death receptor ligands.20 The phosphorylation of other prosurvival signaling proteins was upregulated in the EVLP transplanted lung (compared with the SCS) including, ERK1/2, FAK, PDGFRβ, STAT5/6, and members of the Src family of protein tyrosine kinases Fgr, Hck, and Lck. This broad group of proteins promotes cellular proliferation through increased protein transcription within the nucleus. Increased phosphorylation of 5' adenosine monophosphate-activated protein kinase (AMPK), epidermal growth factor receptor (EGFR), and c-Jun N-terminal kinase (JNK) was also observed. EGFR can inhibit AKT activity and therefore promote apoptosis. AMPK and JNK can be involved in both prosurvival and proapoptotic signaling pathways. Therefore, although AMPK can also inhibit AKT activity and JNK can induce the transcription of proapoptotic genes including p53 and the cleavage of Bid, thereby promoting apoptosis, they can also signal the transcription of proteins involved in cell survival and proliferation.21 Given that there was an increase in the phosphorylation of kinases directly associated with cell survival signaling pathways, it would suggest AMPK, EGFR, and JNK have a prosurvival, rather than proapoptotic role, as increased expression of prosurvival proteins was observed.
Interestingly, EVLP-treated lungs had comparatively higher expression of death receptors (DR) 4 and 5, which drive extrinsic apoptosis pathway activation. There was, however, a concomitant decrease in the intrinsic pathway initiator proteins downstream of DR4 and DR5, including cytochrome C, HTRA2/OMI, and, to a lesser degree, SMAC/Diablo. The decrease in initiator proteins correlated with increases in the antiapoptotic proteins Bcl-2 and PON2, which inhibit their release, and a downregulation of BAX, which normally promotes the opening of the mitochondrial permeability transition pore (MPTP) with subsequent initiator protein release.22 Furthermore, there was an increase in the expression of procaspase 3 in the EVLP group but no increase in its active form, caspase 3. This may be due to inhibition of Apaf-1 by HSP and LIVIN, which prevent downstream activation to caspase 3. Additionally, there was a downregulation of clusterin, which is associated with the clearance of cellular debris and apoptosis. This would suggest increased downstream prosurvival signaling and a reduction in caspase 3 mediated cell death. Therefore, collectively this proteomic profile is indicative of mitochondrial salvage and cell survival in the EVLP transplanted lung tissue, following 24 hours of transplantation
To evaluate this, circulating levels of cell-free gDNA and mtDNA were quantified as surrogate markers of cell death in the recipient circulation posttransplantation. gDNA was selected not only as a marker of cell death but also to distinguish between donor and recipient via identification of Y chromosomal products. mtDNA was selected as a marker of cell membrane rupture and therefore necrosis. However, there was no difference in cell-free gDNA in the recipient circulation of both groups posttransplant. Furthermore, there was no detectable Y chromosome derived DNA, demonstrating that the cell-free gDNA that was present was not donor-derived but of recipient origin. This perhaps is not unsurprising as it may well be too early to observe the systemic changes within the circulation reflective of the signaling pathways within the lung tissue at 24 hours.
Interestingly, immediately following reperfusion of the donor lung, the EVLP group had significantly lower levels of circulating cell-free mtDNA. This is interesting given that cell-free mtDNA drives immune activation via its action as a damage-associated molecular pattern.23 As mtDNA is of bacterial origin, it is high in unmethylated CpG repeats and therefore provokes similar responses to bacterial DNA. Free intracellular fragments can autonomously activate highly specific endosomal toll-like receptors and initiate an amplified immune response, with activation of mitogen activated protein kinase, the release of proinflammatory cytokines, and subsequent neutrophil recruitment and degranulation.23,24 Additionally, cell-free mtDNA acts as a second-messenger molecule, activating the inflammasome, with downstream activation of IL-1β and IL-18.21,24 Outside of transplantation, high levels of circulating cell-free mtDNA are associated with increased intensive care unit mortality, poor clinical outcome following trauma, and increased severity of sepsis.23,25 In the context of transplantation, increased inflammatory signaling leads to augmentation of the recipient immune system as well as activation of donor leukocytes. The subsequent inflammatory infiltration within the graft then promotes graft dysfunction and increased immunogenicity. Therefore, if EVLP is associated with reduced levels of circulating cell-free mtDNA following transplantation, this may attenuate the recipient immune response and improve clinical outcome. This study did not aim to identify the source of cell-free mtDNA or the cause of its release posttransplantation, but we feel delineating the mechanism of release clearly warrants further investigation.
In this experimental transplant model, we report that EVLP alters the signaling profile of the donor lung following transplantation, concomitantly upregulating cell survival and down-regulating proapoptotic and necrotic signaling pathways. Furthermore, EVLP significantly reduces cell-free mtDNA within the peripheral circulation immediately following reperfusion. Collectively, these data indicate that EVLP may exert beneficial effects on the donor lung. If EVLP can reduce inflammation and immunogenicity of the donor lung before transplantation, it may have the potential to reduce PGD and AR posttransplantation and therefore improve long-term clinical outcomes.
1. Chambers DC, Cherikh WS, Harhay MO, et al.; International Society for Heart and Lung Transplantation. The International Thoracic Organ Transplant Registry of the International Society for Heart and Lung Transplantation: thirty-sixth adult lung and heart-lung transplantation Report-2019; focus theme: donor and recipient size match. J Heart Lung Transplant. 2019; 38:1042–1055
2. Diamond JM, Lee JC, Kawut SM, et al.; Lung Transplant Outcomes Group. Clinical risk factors for primary graft dysfunction after lung transplantation. Am J Respir Crit Care Med. 2013; 187:527–534
3. de Perrot M, Liu M, Waddell TK, et al. Ischemia-reperfusion-induced lung injury. Am J Respir Crit Care Med. 2003; 167:490–511
4. Trulock EP. Lung transplantation. Am J Respir Crit Care Med. 1997; 155:789–818
5. Bugge JF. Brain death and its implications for management of the potential organ donor. Acta Anaesthesiol Scand. 2009; 53:1239–1250
6. Kalogeris T, Bao Y, Korthuis RJ. Mitochondrial reactive oxygen species: a double edged sword in ischemia/reperfusion vs preconditioning. Redox Biol. 2014; 2:702–714
7. Fiser SM, Tribble CG, Long SM, et al. Lung transplant reperfusion injury involves pulmonary macrophages and circulating leukocytes in a biphasic response. J Thorac Cardiovasc Surg. 2001; 121:1069–1075
8. Ali JM, Bolton EM, Bradley JA, et al. Allorecognition pathways in transplant rejection and tolerance. Transplantation. 2013; 96:681–688
9. Steen S, Sjöberg T, Pierre L, et al. Transplantation of lungs from a non-heart-beating donor. Lancet. 2001; 357:825–829
10. Wierup P, Haraldsson A, Nilsson F, et al. Ex vivo evaluation of nonacceptable donor lungs. Ann Thorac Surg. 2006; 81:460–466
11. Cypel M, Yeung JC, Liu M, et al. Normothermic ex vivo lung perfusion in clinical lung transplantation. N Engl J Med. 2011; 364:1431–1440
12. Cypel M, Yeung JC, Machuca T, et al. Experience with the first 50 ex vivo lung perfusions in clinical transplantation. J Thorac Cardiovasc Surg. 2012; 144:1200–1206
13. Warnecke G, Van Raemdonck D, Smith MA, et al. Normothermic ex-vivo preservation with the portable organ care system lung device for bilateral lung transplantation (INSPIRE): a randomised, open-label, non-inferiority, phase 3 study. Lancet Respir Med. 2018; 6:357–367
14. Stone JP, Critchley WR, Major T, et al. Altered immunogenicity of donor lungs via removal of passenger leukocytes using ex vivo lung perfusion. Am J Transplant. 2016; 16:33–43
15. Kilkenny C, Browne WJ, Cuthill IC, et al. Improving bioscience research reporting: the ARRIVE guidelines for reporting animal research. Plos Biol. 2010; 8:e1000412
16. Steen S, Kimblad PO, Sjöberg T, et al. Safe lung preservation for twenty-four hours with Perfadex. Ann Thorac Surg. 1994; 57:450–457
17. Steen S, Liao Q, Wierup PN, et al. Transplantation of lungs from non-heart-beating donors after functional assessment ex vivo. Ann Thorac Surg. 2003; 76:244–52. discussion 252
18. Chambers DC, Yusen RD, Cherikh WS, et al.; International Society for Heart and Lung Transplantation. The registry of the international society for heart and lung transplantation: thirty-fourth adult lung and heart-lung transplantation report-2017; focus theme: allograft ischemic time. J Heart Lung Transplant. 2017; 36:1047–1059
19. Avlonitis VS, Fisher AJ, Kirby JA, et al. Pulmonary transplantation: the role of brain death in donor lung injury. Transplantation. 2003; 75:1928–1933
20. Manning BD, Cantley LC. AKT/PKB signaling: navigating downstream. Cell. 2007; 129:1261–1274
21. Zhang W, Liu HT. MAPK signal pathways in the regulation of cell proliferation in mammalian cells. Cell Res. 2002; 12:9–18
22. Danial NN, Korsmeyer SJ. Cell death: critical control points. Cell. 2004; 116:205–219
23. Zhang Q, Raoof M, Chen Y, et al. Circulating mitochondrial DAMPs cause inflammatory responses to injury. Nature. 2010; 464:104–107
24. Schumacker PT, Gillespie MN, Nakahira K, et al. Mitochondria in lung biology and pathology: more than just a powerhouse. Am J Physiol Lung Cell Mol Physiol. 2014; 306:L962–L974
25. Nakahira K, Kyung SY, Rogers AJ, et al. Circulating mitochondrial DNA in patients in the ICU as a marker of mortality: derivation and validation. PLOS Med. 2013; 10:e1001577. discussion e1001577