SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

Pre–liver transplant renal dysfunction is associated with decreased survival following transplantation and is also a prognostic indicator of posttransplant chronic kidney disease. Selection of patients for combined liver/kidney transplantation versus orthotopic liver transplantation alone (OLTa) is often difficult given the lack of a reliable method to predict which patients will have ongoing severe renal dysfunction in the absence of concomitant kidney transplantation. We hypothesized that most patients with pretransplant renal dysfunction (serum creatinine ≥ 1.5 mg/dL for at least 2 weeks prior to and at time of transplant) will not experience a rapid decline in estimated glomerular filtration rates (eGF) post-OLTa to the point of necessitating consideration for kidney transplantation, even in the setting of calcineurin inhibitor–based immunosuppression. We performed a single-center retrospective study of 60 OLTa patients with pretransplant renal dysfunction transplanted between 2000 and 2005. Kaplan-Meier analysis was performed of the time interval to develop eGFR < 20 mL/minute post-OLTa. At OLTa, the mean patient age was 59 years, and median serum creatinine was 1.8 mg/dL; 42% patients were hepatitis C–positive, 32% were diabetic, 38% had kidney dysfunction > 12 weeks, and 5% were receiving hemodialysis. After 36 months median follow-up post-OLTa, only 8 patients (13%) with significant renal dysfunction pre-OLTa achieved eGFR < 20 mL/minute. Patients with pretransplant kidney dysfunction > 12 weeks were at increased risk for eGFR < 20 mL/minute (hazard ratio = 5.3, P = 0.04), a risk that escalated after adjustment for age and serum creatinine at transplant (hazard ratio = 8.9, P = 0.01). Significant predictors of eGFR < 20 mL/minute post-OLTa in this patient cohort were the presence of diabetes and the serum creatinine level at transplant. In conclusion, few patients with preexisting renal dysfunction, especially if <12 weeks duration, experience a significant drop in eGFR post-OLTa. Liver Transpl 14:665–671, 2008. © 2008 AASLD.

Serum creatinine is a key component of the Model for End-Stage Liver Disease (MELD) score, and impaired pretransplantation renal function is known to be a poor prognostic indicator for survival as well as posttransplantation kidney disease.1–4 Furthermore, there are data to support the premise that the duration of pretransplant renal dysfunction (defined as serum creatinine ≥ 1.5 mg/dL) is an important predictor of posttransplant renal function.5 In the current MELD era, an increasing number of patients with renal dysfunction are awaiting orthotopic liver transplantation (OLT)6; this has fuelled debate in the transplant community regarding the optimal therapeutic approach to such patients. Given the scarcity of organ availability, it is important to determine which patients will experience progressive and severe renal dysfunction after orthotopic liver transplantation alone (OLTa) and benefit most from combined liver/kidney transplantation (CLKT). However, the lack of a reliable model to accurately predict renal function post-transplant makes this decision extremely challenging.

Analysis of data from the Scientific Registry of Transplant Recipients indicates that the cumulative incidence of stage 4 [estimated glomerular filtration rate (eGFR) < 30 mL/minute] or 5 chronic kidney disease [eGFR < 15 mL/minute or need for renal replacement therapy (RRT)] post-OLTa is 18% at 5 years.7 The decision for concurrent renal transplantation in patients with hepatic failure is influenced by multiple factors, including the duration and degree of renal dysfunction, the potential for renal recovery post-OLTa, and the impact of dual transplantation on recipient survival.8–10 By conventional criteria, current United Network for Organ Sharing (UNOS) regulations mandate that patients can be listed for kidney transplantation alone only once their eGFR is <20 mL/minute, representing advanced stage 4 chronic kidney disease. Our institutional practice has been to restrict simultaneous renal transplantation to those patients projected to need listing for kidney transplant within 12 months post-OLTa. Against this background, we hypothesized that most OLTa patients with pretransplant kidney dysfunction (serum creatinine ≥ 1.5 mg/dL at transplant, present ≥ 2 weeks) will have sufficient renal reserve to avoid UNOS kidney wait-listing criteria in the first 3 posttransplant years. Based on this hypothesis, the aim of this study was to evaluate renal function post-OLTa over the short and intermediate term following liver transplantation alone in patients with pretransplant kidney dysfunction.

PATIENTS AND METHODS

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

We retrospectively identified a cohort of 60 consecutive OLTa patients with pretransplant renal dysfunction among a total of over 600 liver recipients transplanted at the Hospital of the University of Pennsylvania between March 2000 and August 2005 (that is, both before and after the implementation of the MELD scoring system). Patients included in the study had serum creatinine values of at least 1.5 mg/dL for at least 2 weeks prior to and at time of transplantation. Patient information was obtained from a prospectively maintained database for all transplants performed at our institution. All patients included in our study had at least 1 year of post-OLTa clinical and laboratory follow-up. We also analyzed 13 patients who underwent CLKT at our institution during our study period in order to ensure that our results could be generalized to include these patients. Our study did not include subjects with fulminant hepatic failure and those who had previously undergone liver transplantation. Institutional review board approval was obtained prior to study initiation.

The Hospital of the University of Pennsylvania performs an average of 130 liver transplantations per year. The decision to perform OLTa versus CLKT is made by a team of transplant hepatologists, surgeons, and nephrologists on a case-by-case basis, depending on the etiology, duration, and severity of renal dysfunction.

All serum creatinine values were assessed prior to transplantation. Patients on hemodialysis were assumed to have a serum creatinine of 4.0 mg/dL for the purpose of statistical analysis. The duration of renal dysfunction was defined as the number of weeks prior to transplant during which serum creatinine remained at 1.5 mg/dL or higher. Patients were divided into 2 cohorts based on duration of renal dysfunction: ≤12 weeks or >12 weeks, a cutoff validated by Davis et al.10 The etiology of renal dysfunction was not assessed in our study given the lack of accuracy in identifying the cause of renal dysfunction, particularly in recognition of the scarce number of renal biopsies performed in our cohort of patients.5 Other variables analyzed included the presence of diabetes and/or hypertension, incidence of hepatocellular carcinoma, MELD score, albumin, international normalized ratio, bilirubin, age, race, gender, etiology of liver disease, transjugular intrahepatic portosystemic shunt placement, use of RRT, and the immunosuppression regimen. Laboratory values closest to the time of transplantation were used in our analysis.

Primary outcomes assessed in our study were eGFRs measured with the Modification of Diet in Renal Disease (MDRD) equation [an equation used for the estimation of the glomerular filtration rate (GFR) with serum creatinine in combination with age, sex, and race11] at 6 months, 1 year, 2 years, and 3 years post-OLTa. The MDRD equation was used to estimate GFR as, compared to other formulae, it is a more precise and accurate measure of renal function in liver transplant patients.12 Secondary outcomes were the requirement for chronic hemodialysis post-OLTa and the time to reach eGFR < 20 mL/minute in the subset of patients who experienced a significant decline in renal function. This cutoff was used because it is the UNOS threshold for listing for renal transplantation.

Statistical Methods

Means and standard deviations were reported for parametric variables. For nonparametrically distributed variables, we reported medians and interquartile ranges. Parametric variables were compared with Student t tests; nonparametric variables were compared with rank sum tests, whereas dichotomous variables were compared with Fisher exact tests. Kaplan-Meier analysis was performed of the time to develop eGFR < 20 mL/minute from the time of transplantation. Patients were censored at end of follow-up.

Cox regression was used to assess for any confounding effects on the relationship between the duration of renal dysfunction and the time to reach eGFR < 20 mL/minute.

Potential confounding variables analyzed in the study included serum blood urea nitrogen, creatinine, MELD, albumin, age, race, gender, etiology of liver disease, duration of renal dysfunction, transjugular intrahepatic portosystemic shunt, use of RRT, and the immunosuppression regimen. A P value less than 0.05 was considered significant. Stata 8.1 (Stata Corp, College Station, TX) was the statistical program used for data analysis.

RESULTS

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

Baseline Characteristics

We divided our cohort into 2 groups based on duration of renal dysfunction; 37 patients (62%) had renal dysfunction for ≤12 weeks (mean: 5 weeks), and 23 patients (38%) had renal dysfunction for >12 weeks (mean: 25 weeks). The mean age of our patients prior to OLTa was 59 years. Patients with renal dysfunction for more than 12 weeks were older than those with shorter duration of renal insufficiency (mean: 63.2 years versus 55.5 years, P < 0.001). There was no gender difference between the 2 cohorts; 42% of our patients were hepatitis C–infected, and 18% had alcoholic cirrhosis. There was no statistically significant difference in the etiology of liver disease between the 2 groups (Table 1).

Table 1. Characteristics of Patients with Long and Short Durations of Renal Dysfunction Prior to Transplantation
VariableDuration of Renal Dysfunction ≤ 12 Weeks (n = 37)*Duration of Renal Dysfunction > 12 Weeks (n = 23)*P Value
  • Abbreviations: HCC, hepatocellular carcinoma; HCV, hepatitis C virus; INR, international normalized ratio; MELD, Model for End-Stage Liver Disease; N/A, not applicable; NASH, nonalcoholic steatohepatitis; OLT, orthotopic liver transplantation; TIPS, transjugular intrahepatic portosystemic shunt.

  • *

    Data are presented as mean ± standard deviation, n (%), or median (interquartile range).

Duration of renal dysfunction (weeks)5 (3, 7)25 (18, 43)N/A
Age (years)55.5 ± 8.363.2 ± 8.2<0.001
Male gender27 (73%)19 (83%)0.53
Caucasian race32 (86%)22 (96%)0.39
Etiology   
 HCV ± alcohol16 (43%)9 (39%)0.90
 Alcohol6 (16%)5 (22%) 
 NASH/cryptogenic8 (22%)6 (26%) 
 Other7 (19%)3 (13%) 
HCC8 (22%)5 (23%)1.00
Diabetes11 (30%)8 (35%)0.78
Hypertension13 (35%)6 (26%)0.57
Hemodialysis prior to OLT2 (5%)1 (4%)1.00
TIPS performed prior to OLT5 (14%)3 (13%)1.00
Creatinine (mg/dL)1.8 (1.6, 2.2)1.7 (1.5, 2.6)0.58
INR1.8 (1.4, 2.1)1.5 (1.3, 1.7)0.01
Bilirubin (mg/dL)3.5 (2.4, 5.5)2.1 (1.4, 4)0.08
Albumin (g/dL)2.5 ± 0.62.6 ± 0.50.38
MELD24.2 ± 7.220.5 ± 5.80.04

Median serum creatinine at time of transplant was 1.8 mg/dL; there was no difference between the 2 cohorts in the serum creatinine value at time of transplant (1.8 versus 1.7 mg/dL, P = 0.58). Thirty-two percent of the patients had diabetes mellitus, and 5% were on hemodialysis prior to OLTa. The MELD scores between the 2 cohorts were significantly different (24.2 versus 20.5, P = 0.04); given that the serum creatinine values were similar in the 2 cohorts, the difference in MELD scores was driven primarily by significant differences in international normalized ratio levels between the 2 cohorts (1.8 versus 1.5, P = 0.01).

Comparing patients who underwent CLKT versus OLTa to ensure the generalizability of our results, we found that CLKT patients were younger than those undergoing OLTa (mean age: 52 versus 58.2 years, P = 0.03); significantly more CLKT patients were on hemodialysis prior to transplantation compared to OLTa patients. The median duration of renal dysfunction was longer in CLKT patients (16 versus 8 weeks, P = 0.03), and the median serum creatinine value at transplant was higher in the CLKT cohort (4 versus 1.8 mg/dL, P < 0.001). The mean albumin in CLKT patients was also lower than that in OLTa patients (Table 2). The etiology of renal dysfunction in CLKT patients was as follows: long duration of hepatorenal syndrome (3 patients), hypertensive nephropathy (1 patient), nonsteroidal anti-inflammatory drug nephropathy (1 patient), renal cell carcinoma (2 patients), membranoproliferative glomerulonephritis (3 patients), polycystic kidney disease (1 patient), lupus nephritis/cyclosporine toxicity (1 patient), and unknown (1 patient).

Table 2. Comparison of Characteristics Between CKLT and OLTa Patients
VariableCKLT (n = 13)*OLTa (n = 60)*P Value
  • Abbreviations: CLKT, combined liver/kidney transplantation; HCV, hepatitis C virus; HD, hemodialysis; INR, international normalized ratio; NASH, nonalcoholic steatohepatitis; OLTa, orthotopic liver transplantation alone; TIPS, transjugular intrahepatic portosystemic shunt.

  • *

    Data are presented as mean ± standard deviation, n (%), or median (interquartile range).

Age (years)52 ± 10.358.4 ± 9.00.03
Male gender11 (85%)46 (77%)0.72
HCV ± alcohol9 (69%)25 (42%) 
Alcohol2 (15%)11 (18%)0.18
NASH/cryptogenic0 (0%)14 (23%) 
Other2 (15%)10 (17%) 
Diabetes7 (54%)19 (32%)0.20
Hypertension3 (23%)19 (32%) 
HD prior to transplant7 (54%)3 (5%)<0.001
Duration of renal dysfunction (weeks)16 (6, 72)8 (4,21)0.03
Serum creatinine (mg/dL)4 (2.7, 4)1.8 (1.5, 2.3)<0.001
TIPS performed prior to transplant2 (15%)8 (13%)1.00
INR1.6 (1.4, 1.8)1.7 (1.4, 2.1)0.80
Bilirubin (mg/dL)3.8 (1.4, 5.8)3.1 (1.5, 5.1)0.15
Albumin (g/dL)2.1 (1.9, 2.5)2.5 (2.2, 2.9)0.01

There was no statistically significant difference in the immunosuppression regimen (defined as the medication used for the majority of the first year post-OLTa) in our 2 cohorts separated by duration of pretransplant renal dysfunction. The vast majority of patients were maintained on tacrolimus post-OLTa (Table 3); however, levels were not specifically assessed in this study as patients were maintained on a protocol regimen and medications were dose-adjusted to avoid calcineurin inhibitor–induced nephrotoxicity.

Table 3. Immunosuppression Regimen and Renal Outcomes After OLT for Patients with Long and Short Durations of Renal Dysfunction Prior to OLT
VariableDuration of Renal Dysfunction ≤ 12 Weeks (n = 37)*Duration of Renal Dysfunction > 12 Weeks (n = 23)*P Value
  • Abbreviations: eGFR, estimated glomerular filtration rate; MMF, mycophenolate mofetil; OLT, orthotopic liver transplantation; OLTa, orthotopic liver transplantation alone.

  • *

    Data are presented as n (%) or median (interquartile range).

Prednisone19 (51%)16 (70%)0.19
MMF10 (27%)9 (39%)0.40
Tacrolimus31 (84%)19 (83%)1.00
Sirolimus9 (24%)3 (13%)0.34
eGFR 6 months after OLTa46 (36, 61)40 (34, 51)0.39
eGFR 12 months after OLTa50 (36, 62)45 (32, 54)0.19
eGFR 24 months after OLTa57 (40, 66)45 (38, 58)0.21
eGFR 36 months after OLTa58 (40, 70)49 (34, 61)0.42

Posttransplant Kidney Function

The median eGFR measured at 6 months, 1 year, 2 years, and 3 years post-OLTa was not significantly different between our 2 cohorts (Table 3).

After median duration follow-up of 36 months post-OLTa, only 8 patients (13%) with significant renal dysfunction pre-OLTa achieved eGFR < 20 mL/minute. Of these 8 patients, 6 had kidney dysfunction for more than 12 weeks prior to OLTa. All 6 patients were either diabetic or had a serum creatinine value > 2 mg/dL at the time of transplantation. None of these patients received kidney transplantation during our study period; 6 patients needed to start chronic hemodialysis a mean of 16 months post-OLTa. The Kaplan-Meier curve (Fig. 1) reveals the fraction of patients who maintained eGFR > 20 mL/minute at 36 months: 35 out of 37 patients (95%) with renal dysfunction ≤ 12 weeks and 17 out of 23 patients (74%) with renal dysfunction > 12 weeks.

thumbnail image

Figure 1. Patients with GFR > 20 after OLT.

Download figure to PowerPoint

The hazard ratio for prediction of eGFR < 20 mL/minute for patients with renal dysfunction for greater than 12 weeks was 5.3 (1.1–26.1; P = 0.04) compared to patients with duration of renal dysfunction < 12 weeks. After adjustment for age and serum creatinine value at time of transplant on multivariate analysis, the hazard ratio was 8.9 (1.6–49.4; P = 0.01).

Among the 23 patients with duration of renal dysfunction > 12 weeks, the only significant predictors of eGFR < 20 mL/minute post-OLTa were the presence of diabetes and the serum creatinine value at time of transplant. Five of the 8 diabetics in this cohort achieved eGFR < 20 mL/minute. Four of the 8 patients with serum creatinine > 2 mg/dL at time of transplant achieved eGFR < 20 mL/minute.

DISCUSSION

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

Given the current debate on the utility of CLKT due to the increasing number of patients with renal dysfunction awaiting OLT, our data are reassuring, showing that the vast majority of patients with pretransplant renal insufficiency do not develop advanced stage 4 or 5 chronic kidney disease post-OLTa, even in the setting of calcineurin inhibitor–based immunosuppression, a known nephrotoxic risk factor.13 In our study, only 13% of patients with significant renal dysfunction prior to OLTa reached an eGFR of <20 mL/minute post-OLTa at 36 months' median follow-up. Severe renal dysfunction was particularly uncommon among patients whose pretransplant serum creatinine was elevated for less than 12 weeks; only 2 patients (5%) within this cohort reached this endpoint. Our data correlate with an earlier study done at our institution validating that the duration of pretransplantation serum creatinine elevation predicts post-OLTa renal dysfunction.5 We did not stratify our patients on the basis of etiology of renal dysfunction, given the lack of accuracy in classifying kidney disease in the absence of renal biopsies and the concern for misclassification bias, particularly with respect to the diagnosis of hepatorenal syndrome. Furthermore, in an earlier study performed by Campbell et al.5 at our institution, etiology of renal dysfunction was not significantly associated with posttransplant renal outcomes, likely secondary to inaccurate classification of renal disease in these patients.

In terms of comparisons between our study and the Campbell study, we believe that, although there are some similarities between the 2 studies, there are important differences. There are still no definitive criteria for determining who should get OLTa versus CKLT, and all studies to date, including ours, are retrospective. The Campbell study used RRT at 6 and 12 months as the outcome to try to identify CKLT candidates; however, we have since increasingly recognized that this may not be the most appropriate or clinically relevant endpoint for the following reasons: (1) need for RRT at 6 or 12 months is too restrictive and does not capture patients with slightly lesser degrees of advanced kidney disease who nevertheless require RRT preparation and could also theoretically benefit from CKLT; (2) our institutional practice has been to restrict simultaneous renal transplantation to those patients projected to need listing for kidney transplant within 12 months post-OLTa; and (3) equally important, the current study provides extended, rather than short-term, follow-up of our original cohort to 36 months.

With these modified outcome measures, we have also now identified that, among patients with kidney dysfunction for more than 12 weeks prior to OLTa, pretransplant serum creatinine > 2 mg/dL and the presence of preexisting diabetes are significant predictors of poor posttransplant renal outcomes (eGFR < 20 mL/minute within 3 years post-transplant). We consider the confirmation of our previous observation with extended follow-up and use of a broader outcome measure to be reassuring in terms of using longer duration of pre–liver transplant kidney dysfunction (serum creatinine > 2.0 for more than 12 weeks) as a criterion for CLKT. These findings are particularly pertinent given that rates of CLKT have continued to grow, while simultaneously the waiting list for kidney transplantation now exceeds 70,000, underscoring the need for more discriminatory selection criteria for the combined transplant.

With respect to patient selection, our study included an additional 22 patients transplanted between August 2003 and August 2005. Additionally, 18 of the 56 patients included in the earlier study had serum creatinine ≥ 1.5 mg/dL for less than 2 weeks; hence, they did not meet our study's inclusion criteria. The median serum creatinine in our study was not significantly different from the earlier study (1.8 versus 1.7 mg/dL); however, the duration of renal dysfunction in our patients was longer (median: 8 versus 2.7 weeks).5

We selected a pretransplant serum creatinine cutoff of 1.5 mg/dL to define kidney dysfunction as this value has been validated in other studies, and it also signifies more advanced kidney disease in patients with end-stage liver disease than in patients without cirrhosis.14, 15 Post-OLTa, an eGFR of <20 mL/minute was chosen for our primary endpoint for 2 reasons: (1) it represents advanced stage 4 chronic kidney disease, a level of function at which national guidelines recommend initiation of preparation for long-term RRT, and (2) it is the UNOS threshold for placement on the waiting list for kidney transplantation. We used the MDRD equation to estimate GFR as this has been validated to be more precise in liver transplant patients than other renal formulae.12 At the same time, we recognize that despite its greater precision, the MDRD equation actually underestimates GFR measured by the gold standard iothalamate clearance.12 The implications of this for our study are that the small number of our patients who reached an eGFR of <20 mL/minute may, if anything, represent an overestimation.

Our multivariate analysis revealed that, after adjustment for age and serum creatinine at the time of transplant, post-OLT eGFR < 20 mL/minute was predicted by the presence of pretransplant kidney dysfunction for more than 12 weeks (hazard ratio = 8.9, confidence interval = 1.6–49.4, P = 0.01). Six patients in our study needed chronic RRT after OLTa; all of these patients had pretransplant renal dysfunction for >12 weeks.

Among the subset of patients with longer duration of renal dysfunction, the only significant predictors of eGFR < 20 mL/minute post-OLTa were the presence of diabetes prior to transplant and the serum creatinine value at time of transplant, factors strongly associated with chronic kidney disease. Other variables including MELD score and etiology of liver disease, which have been associated with postoperative renal dysfunction in other studies,5, 16 were not found to be significant predictors of renal dysfunction in our study. Eighteen of our 60 total patients were transplanted prior to the implementation of the MELD scoring system, and the proportions of patients with short-duration and long-duration renal dysfunction were similar in pre-MELD and post-MELD transplant patients.

Our data further validate the notion that the extent and duration of kidney disease are important determinants of severe renal insufficiency post-OLTa. Furthermore, the presence of diabetes is a critical component in determining renal function after liver transplantation.

We recognize additional limitations to our current study. First, consistent with most other studies in this area, our study comprised a retrospective design. Second, because the median serum creatinine value in our cohort were not as high as the level in CLKT patients, the possibility of selection bias cannot be excluded. However, a major finding of both our study and that of Campbell is that the duration of renal dysfunction is far more important than the actual creatinine value in predicting adverse kidney outcomes. The median duration of renal dysfunction in our 13 CLKT patients was 18.1 weeks; hence, on the basis of our hypothesis, they would have been considered for dual organ transplantation anyway. Third, patients who, within the first 6 posttransplant months, had unfavorable outcomes or were lost to follow-up were excluded from the analysis. Although it is likely that patients with unfavorable early outcomes had severe perioperative kidney injury, this could almost certainly have occurred in the setting of other overwhelming complications such as liver allograft dysfunction, sepsis, hemodynamic instability, and multiorgan system failure. Fourth, patients with serum creatinines > 1.5 mg/dL for less than 2 weeks pre-transplant were omitted in order to exclude patients who may have experienced only a transient, easily reversible renal insult such as volume contraction. Finally, because our current standard of nephrology practice is to generally not perform kidney biopsies in patients with decompensated liver failure, we cannot rule out histological evidence of chronic kidney disease such as diabetic nephropathy that was masked by the clinical picture and pretransplant acute or subacute kidney injury. However, similar criteria were applied to all patients in determining whether or not they should receive a CKLT.

In summary, our data suggest that most OLTa patients with pretransplant kidney dysfunction do not experience a rapid progression to advanced stage 4 or 5 chronic kidney disease after transplantation. Among those with longer duration of renal dysfunction prior to transplantation, the presence of diabetes and the serum creatinine value at time of transplant may help identify a small subset of patients who could be considered for CLKT as opposed to liver transplantation alone. Furthermore, CLKT is seldom warranted for patients with pretransplant kidney dysfunction present for less than 12 weeks. Larger observational studies are needed, particularly those including cost analyses, in order to justify the approach of liver transplantation alone versus CLKT in those with longer duration of kidney dysfunction prior to liver transplantation.

REFERENCES

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES