Introduction

Anorexia nervosa is a rare disease. Very few studies report the incidence of anorexia nervosa in the general population but anorexia nervosa is mainly observed in adolescents and young female. The heterogenous populations studied (ambulatory or hospitalized) and the diversity of diagnostic tools used to perform diagnosis can explain the differences between studies. In a French cohort, the prevalence of anorexia nervosa was 1.8% in a female population aged between 18 and 25 years. It decreased to 1.4% in a female population aged 25–45 years. No incidence data were available [1]. The prevalence of anorexia nervosa in a representative sample of French adolescents was 0.5% for girls and 0.03% for boys [2]. Two subtypes of anorexia nervosa (restricting and binge eating/purging) are defined in the Diagnostic and Statistical Manual of Mental Disorders (Fourth Edition, Text Revision) (DSM-IV-TR) [3]. Rigaud et al. followed 484 French patients with anorexia nervosa for 13.5 years, and found 6% poor outcome (defined as BMI 16.5–17.5 kg/m2 or abnormal eating behavior), 6% severe outcome (BMI < 16.5 kg/m2 and/or at least one binge/purging episode per day or evolution to bulimia nervosa) and a mortality rate of 1.2% [4].

Furthermore, anorexia is a severe disease associated with medical complications and a 5-year mortality rate of about 3% [5, 6]. Little known kidney injuries complications including electrolyte abnormalities and acute and chronic renal failure can lead to end-stage renal disease (ESRD) [7]. In France, in 2011, according to the French Agence de la Biomedecine, the overall crude prevalence of patients receiving renal replacement therapy (dialysis or kidney graft) was 1091 per million inhabitants. In a prospective 21-year follow-up study of 84 patients with anorexia nervosa, Zipfel et al. found that 5.2% of patients required hemodialysis for ESRD [8]. Between 2006 and 2011, French patients on dialysis increase from about 25,774 to 39,600 and patients living with a functional graft from about 19,491 to 31,100. Transplant from a living donor increases gradually in France, accounting for about 2.9 and 10.1% of kidney transplants, respectively, in 1989 and in 2011 according to the French Agence de la Biomedecine.

The appropriate care of anorexia nervosa associated with chronic kidney disease (CKD) presents a challenge for nephrologists. In case of anorexia nervosa patients, the best treatment modality is not known but there are no guidelines to exclude this patient from kidney transplantation. It is not recommended to start a pre-transplant assessment while psychiatric disorder is still ongoing. A specific psychiatric care before being on the waiting list for renal graft is required for these patients. To our knowledge, there are only one report of kidney transplantation and few reports in dialysis. Osório et al. reported a 41-year-old woman with long-term untreated anorexia nervosa-binge eating/purge type who developed ESRD requiring maintenance peritoneal dialysis. This woman died of septicemia due to peritonitis [9]. Woodside et al. also reported a case of a 39-year-old woman with a 25-year history of anorexia nervosa–binge eating/purge who was successfully weight restored while on hemodialysis [10]. Kidney transplantation is usually the treatment of choice for ESRD in the general population. Both eating disorders and uremic syndrome related to chronic renal failure lead to important nutritional alterations. Several studies have identified significant disadvantages of an altered nutritional status in terms of kidney transplant survival [11,12,13]. There are no specific published data addressing graft survival in anorexia nervosa patients, who are most of the time young women candidate for renal transplantation.

Here, we report a multicenter, retrospective, case–control study that assessed graft survival in patients with anorexia nervosa during a 5-year follow-up and compared to graft survival in kidney transplant-matched patients with either low BMI or with normal BMI.

Patients and methods

Source of patients and medical record data

We asked 35 renal transplant units in France to retrospectively identify kidney transplant recipients who were diagnosed with anorexia nervosa at time of transplant. As controls, we identified adult transplant recipients seen between 1989 and 2011 with a BMI at the time of transplantation of less than 24.9 kg/m2. Patients with a BMI > 25 kg/m2 were excluded. Clinical data were collected from Bordeaux R@N medical record database (approved by Comité National de l’Informatique et des Libertés CNIL No 1357164) and from the multicentric, observational and prospective DIVAT cohort (Données Informatisées et VAlidées en Transplantation; https://www.divat.fr, approved by the Comité National de l’Informatique et des Libertés CNIL, No. 914184). DIVAT and R@N database are two electronic devices used as medical records in the routine care of kidney-transplanted patients. DIVAT database exists since 1995 and is used for several publications [14]. For the other centers, data were collected from medical records.

Patients were divided into three groups: anorexia nervosa, low BMI, and normal BMI. Post-transplant patients were followed for 5  years or until graft loss or death. For each case of anorexia nervosa, we randomly included at least three patients in each control group matched for age, center, and transplantation date.

Additional inclusion criteria

Anorexia nervosa group

The inclusion criteria were kidney transplant recipient from a deceased donor between 1989 and 2011, and a diagnosis of anorexia nervosa as defined by the DSM-IV-TR criteria at the time of transplant. Medical records were carefully examined to verify that patients met the DSM-IV-TR criteria [3]

Control groups

Inclusion criteria were kidney transplant recipient from a deceased donor between 1989 and 2011, and BMI less than 24.9 kg/m2 at the time of transplantation.

Controls were divided into low BMI or normal BMI groups according to the World Health Organization classification: low (BMI < 18.5 kg/m2), normal (BMI 18.5–24.9 kg/m2) [15].

Exclusion criteria

Exclusion criteria for both anorexia nervosa and control groups were recipients with BMI below 18.5 kg/m2 that could not be explained by a idiopathic thinness (major depressive disorder or other major psychiatric disease at the time of grafting, chronic infections, gastrointestinal disease, chronic inflammatory disease, advanced liver disease responsible for weight loss, patients with early graft loss (graft failure within 30 days of transplantation), patients who received a living-donor kidney. As all the anorexia nervosa patients were female, no male were included in the control group to get gender matching.

The primary end point was graft survival over a 5-year follow-up. We defined graft survival as the time between the date of transplantation and the date of graft failure or death of the patient.

Secondary endpoints included analysis of BMI course and renal allograft function (excluding graft losses) at 5 years, and the occurrence of other complications such as chronic infections, malignancies, cardiovascular disease (defined as coronary artery, cerebrovascular, or peripheral arterial disease), bone disease (defined as osteopenia or osteoporosis), and diabetes.

Definition of clinical parameters during the 5-year follow-up

Delayed graft function was defined as the need for dialysis within 7 days post-transplantation. Acute rejection episodes had to be biopsy proven.

Estimated glomerular filtration rate (GFR) was calculated using the modification of diet in renal disease (MDRD) formula [16]. For patients who returned to dialysis during the follow-up, GFR was estimated at the time of the return to dialysis. Cytomegalovirus (CMV) infection was defined by a positive DNAemia using a whole-blood quantitative CMV real-time polymerase chain reaction assay.

Immunosuppressive regimen

As induction therapy, all patients received methylprednisolone and either basiliximab or OKT3 monoclonal antibody or anti-thymocyte globulin. A triple immunosuppressive regimen with either tacrolimus or cyclosporine A concomitantly with prednisone and either mycophenolate mofetil or azathioprine was used for maintenance.

Statistical analysis

Data are shown as means ± SDs, medians (lower and upper extremities) and percentages, as appropriate. Demographic and clinical data were compared among groups with the Pearson’s chi-squared test for categorical variables and analysis of variance for continuous variables. A Kaplan–Meier survival curve was constructed for graft survival. Association between clinical variables and graft failure was estimated using univariate analysis. Multivariate analysis was performed using Cox’s regression model. The multivariate model incorporated a backward and stepwise elimination method using variables with p less than 0.25 from the univariate analysis. Separate analyses were carried out on subjects with anorexia nervosa. Clinical data and eGFR at 3, 6, and 12 months were compared between patients with or without graft loss in the anorexia group by Wilcoxon non-parametric tests and Fisher test. Statistical analysis was performed using STATISTICA version 10 (StatSoft Inc, Tulsa, USA) and R 2.1.1 (R Development Core team 2004). p < 0.05 was considered statistically significant.

Results

Main characteristics of the patient groups

Eighteen transplantation centers answered the questionnaire. A total of 137 subjects were included in this study (Fig. 1). 30 patients with anorexia nervosa were identified: 5 were excluded, as they did not reach the DSM-IV-TR criteria of anorexia nervosa; 2 were excluded because they received a living donor and could not be matched with low BMI recipients in the same center, and 4 because of confounding factors (gastrointestinal disease, graft loss within 30 days). Only 2 individuals out of 19 anorexia nervosa patients were of restricting type. No patient met diagnostic criteria for bulimia nervosa at follow-up. Finally, patients from 4 centers were enrolled in the study, 19 with anorexia nervosa, 59 with low BMI, 59 with normal BMI (Necker, Paris, n = 8/24/26; Bordeaux, n = 4/13/12; Nantes, n = 5/16/15; Grenoble, n = 2/6/6) (Fig. 1).

Fig. 1
figure 1

Study flowchart

Full size image

Baseline patients and transplantation characteristics are shown in Table 1. At baseline, there were no differences between mean BMI in the anorexia group and the low BMI group (16.7 ± 1.6 kg/m2 vs. 17.0 ± 1.1 kg/m2, respectively, p = NS). As expected, mean BMI in the anorexia group was lower than mean BMI in the normal group (16.7 ± 1.6 kg/m2 vs. 21.4 ± 1.5 kg/m2, p < 0.001).

Table 1 Baseline characteristics of the study population
Full size table

Graft survival in the anorexia nervosa group compared to control groups (Fig. 2)

Fig. 2
figure 2

Graft survival in the anorexia nervosa group compared to control groups

Full size image

Graft loss had occurred in 7 of 19 anorexia nervosa patients within 5 years after transplantation (Fig. 2). The cause of graft loss was identified as rejection in two cases and as recurrence of initial disease in one case. Cause of rejection was not determined in four cases since no biopsy was performed. Graft survival was significantly reduced in anorexia nervosa patients (p = 0.01) compared to low and normal BMI. There were no differences in graft survival between low and normal BMI patients (p = 0.12).

In multivariate analysis, anorexia nervosa (HR 5.53 [95% CI 3.37–8.98], p = 0.005) and delayed graft function (HR 2.63 [95% CI 1.26–5.53], p = 0.01) were associated with a significantly increased risk of graft loss (Table 2).

Table 2 Univariate and multivariate logistic regression by Cox model
Full size table

Age at transplantation, duration of dialysis, history of smoking, history of hypertension, HLA-mismatches, expanded criteria donor (ECD), cold ischemia duration, biopsy proven acute rejection, and CMV infection had no significant impact on graft survival as estimated by the Cox model.

No deaths occurred in the normal BMI group. In the anorexia nervosa group, three deaths (16%) occurred during the first year following return to dialysis, and five deaths (8%) occurred in the low BMI group among patients with a functional graft (p = 0.16).

Evolution of BMI in the different groups

In the anorexia nervosa group, BMI increased during the first 5 months after transplantation (16.7 ± 1.6 kg/m2 to 18.2 ± 2.1 kg/m2) but remained stable thereafter. In the low- and normal-BMI groups, BMI increased gradually throughout the 5 years of follow-up after transplantation (low-BMI group: 17.0 ± 1.1 kg/m2 to 19.2 ± 2.4 kg/m2; normal-BMI group: 21.4 ± 1.5 kg/m2 to 23.4 ± 3.7 kg/m2). We did not observe any correlations between changes in weight at 1 year and kidney graft survival (p = 0.56).

Evolution of kidney function in the different groups

There was no significant difference among the three groups in estimated GFR (p = 0.48) at the 5-year post-transplantation follow-up. In the anorexia nervosa group, the occurrence of acute kidney injury was higher in patients who lost their graft (86%, n = 6) versus those who did not (75%, n = 9) (p < 0.001).

Occurrence of complications

There were more cardiovascular events after kidney transplantation in the anorexia group (37%) than in low (6%) and normal BMI groups (7%) (p = 0.001).

However, there was no difference in delayed graft function (p = 0.19), biopsy proven acute rejection (p = 0.16), CMV infection (p = 0.35), bacterial infection (p = 0.87), cancer (p = 0.68), bone disease (p = 0.66), diabetes (p = 0.64), or psychiatric follow-up (p = 0.06).

Discussion

In this retrospective multicenter study, we found that anorexia nervosa was a strong independent risk factor associated with outcomes after kidney transplantation. In the multivariate model, only anorexia nervosa and DGF, as previously reported [17], were significantly associated with poor graft survival.

The best modality of renal replacement therapy in this population has not yet been defined. There is only one case report that discusses the complexities of managing kidney transplantation in these patients [18]. Hampton et al. described a case of a second kidney transplant in a 44-year-old woman who had suffered from anorexia nervosa since she was 18 years old. Complications occurred during the first post-operative week caused by severe fluid retention. She became obsessed with her fluid status and compensated by controlling her fluid intake. This complication resulted in a recurrence of the eating disorder related to change in body image.

The poor graft survival associated with anorexia nervosa that we observed is probably multifactorial. Anorexia nervosa is associated with low BMI and nutritional alterations. Several studies have associated low BMI with shorter graft and patient survival [11,12,13, 19,20,21,22]. Sezer et al. demonstrated that low BMI (< 19 kg/m2) was associated with chronic rejection (73.9 versus 20.7%, p < 0.001) and graft loss (73.9 versus 31.5%, p < 0.001) at 5-year post-transplantation, but was not associated with acute rejection (p > 0.05) [21]. Analysis of 5552 patients from the ANZDATA registry showed that being underweight (defined as a BMI < 18.5 kg/m2) was associated with a 27% increased risk for death-censored graft loss [11]. In contrast, we did not find any difference in graft survival between low and normal BMI control groups probably because of the small sample size. Poor nutritional status has a negative impact on patient survival [23,24,25]. Chang et al. found that more than 5% weight loss during transplantation is associated with an increased mortality rate [26]. Deaths were from cardiovascular disease (35%) and cancer (35%). In our anorexia nervosa population, we showed that cardiovascular events before and during the graft period were more frequent than in the other groups. The worse prognostic we noticed in the anorectic group could be related to other factors independent of weight. All anorexic patients were on dialysis before transplantation, while 10% of control patients had a pre-emptive transplant, which is known to improve long-term kidney transplant function and survival. All renal complications related to eating disorders (acute kidney injury, hypokalemia, hyponatremia, hypocalcemia, hypomagnesemia, hypophosphatemia and renal stones) can also occur after transplantation. In this regard, we noticed that patients in our anorexia nervosa population who lost their graft exhibited more frequent acute kidney injury. Furthermore, physicians may be embarrassed by drug nephrotoxicity due to alterations in pharmacokinetics. Serum lipophilic drug concentration as cyclosporin may be higher in underweight patients, leading to a greater nephrotoxicity. We could not study this issue, as data on drug levels and compliance were not collected in our cohort. In the absence of matching between anorexia nervosa group and low-BMI group, a possible bias may also explain lower kidney graft survival in the anorexia nervosa group: there were no pre-emptive kidney transplantation in the anorexia nervosa group whereas five were performed preemptively in the low BMI group.

We observed that, in contrast to the two control groups, anorexia nervosa was associated with a different course of BMI after successful kidney transplantation, as BMI remained low during the 5-year post-transplantation follow-up, whereas patients with low or normal BMI showed a continuous increase in BMI. Usually, nutritional status improves after renal transplantation and is accompanied with an increase in body weight [27,28,29]. For example, two studies previously found a significant (13.5%) increase in weight after a 5-year follow-up, which included an increase in both fat and lean mass [30, 31]. Several factors can influence weight after renal transplantation including steroid administration, nutritional behavior, and higher grade uremia [30, 32,33,34]. It is likely that in the anorexia nervosa group, nutritional behavior (food restriction, binge/eating purging) is the main mechanism responsible for the lack of body weight increase compared to the overall kidney transplant population.

There are limitations to our study, which should be considered when interpreting the data. We did not find an impact of anorexia nervosa on eGFR evolution calculated with MDRD formula at 5-year follow-up probably because of the small sample size. But this raises the question of reliability of eGFR calculated with MDRD formula in low BMI and anorexia patients. In fact, serum creatinine is affected by muscle mass and none of the eGFR prediction equations has been validated in low BMI cases [35, 36]. To overcome the limitations due to the retrospective character of our study and heterogeneity in the immunosuppressive regimens, we matched the three groups according the year of kidney transplantation. All renal units in France were interviewed. More than 50% of them answered the questionnaire. These centers represent 65% of kidney transplantations performed in France, although the sample size is small, it can be considered as representative of the population of French transplanted patients meaning that the prevalence of anorexia nervosa in kidney transplantation is very low. So, to obtain 19 patients, the recruitment of the population extended over 22 years. To exclude the effect of acute events such as surgical complications and only to document long-term effects, patients who exhibited loss of transplant function within the first month after transplantation were not included in this analysis. Some data would be useful to understand the poor outcome in transplantation: body composition, measured GFR, treatment adherence and blood levels of immunosuppressive agents and level of severity of anorexia nervosa in our population. We know that the level of severity is based, for adults, on current body mass index. For our group, the mean body mass index was 16.7 kg/m2 corresponding to a moderate level of severity. But the level of severity depends also on clinical symptoms, the degree of functional disability and the need for supervision, and data we did not collect retrospectively. 89.5% of the anorexia patients where binge-purging type (50% in the overall French anorexia population). Binge-purging type anorexia nervosa is more frequently associated with metabolic abnormalities that can lead to chronic kidney disease. This may explain the over representation of this disorder in our anorexia nervosa group.

It would be valuable to assess morbidity and mortality of this specific population in extra renal therapies. In conclusion, anorexia nervosa is associated with poor prognosis after kidney transplantation. The best treatment in case of ESRD in this population is largely unknown. In clinical practice, kidney graft should be an option but a psychological impact is possible and can compromise the patient’s quality of life. Careful pre- and post-transplant screening and appropriate care, with specific attention to nutritional behavior and weight course, are indicated. Initial assessment and psychological support are necessary because the psychological impact of the kidney transplant can induce major depressive disorder, anxiety disorder, substance use disorder or body dysmorphic disorder [37]. Psychological approach must evaluate the risk of body dysmorphic disorder due to the surgery or drug side effects. A multidisciplinary approach to care, including psychiatrists and clinicians, should be pursued.