Abstract
We investigated the relationship between preoperative comorbidity and postoperative survival after intestinal transplantation. Each patient received a score for preoperative comorbidity. Each comorbidity was given a score based on the degree it impaired function (score range 0–3). A total score was derived from the summation of individual comorbidity scores. Patients (72 adults (M : F, 33 : 39)) received an isolated intestinal graft (27) or a cluster graft (45). Mean (standard deviation) survival was 1501 (1444) days. The Kaplan-Meier analysis revealed a significant inverse association between survival and comorbidity score (logrank test for trend, ). Patients grouped into comorbidity scores of 0 and 1, 2 and 3, 4 and 5, 6, and above had hazard ratios (95% confidence intervals) for death (compared to group 0 + 1), which increased with comorbidity scores: 1.945 (0.7622–5.816), 5.075 (3.314–36.17), and 13.77 (463.3–120100), respectively, (). Receiver-operator curves at 1, 3, 5, and 10 years postoperative had “C” statistics of 0.88, 0.85, 0.88, and 0.92, respectively. When evaluating patients for transplantation, the degree of comorbidity should be considered as a major factor influencing postoperative survival.
1. Introduction
Over the last decade, intestinal transplantation has become an established treatment modality in the management of intestinal failure. The preoperative status of patients has been found to influence the outcome of surgery [1], and scores developed to semiquantify this [2, 3] have been used in routine practice to facilitate preoperative risk assessment. There has been considerable improvement in postoperative survival over the last 20 years. Much of this improvement has arisen from advances in immunosuppressive regimes and better preoperative preparation and postoperative care of patients [4]. Although survival after transplantation remains inferior to that on parenteral nutrition (PN), the gap is closing, and as a consequence patients are being considered for transplantation at an earlier stage. Potential improvement in quality of life is also now being factored in to the decision regarding a patient’s suitability for transplantation. It has therefore become very important to accurately assess patient’s individual survival chance. In our routine clinical practice, we have considered preoperative comorbidity to have a negative influence on postoperative survival [3]. We have semiquantified comorbidity for each patient and compared this to postoperative survival to determine the relationship.
2. Methods
Preoperative comorbidity was transformed into a numerical score to allow service analysis to take place. This work was conducted according to the requirements of the institution. The score was compared to subsequent survival. Comorbidity factors included were loss of venous access and impairment of organs or systems not corrected by transplantation. Each factor was scored 0–3. A score of 3 indicated a functional abnormality not fully corrected by treatment of a severity approaching a contraindication for transplantation. A score of 1 describes a characteristic which does not present an immediate risk but which may develop into a risk factor in the future. A score of 2 indicates an abnormality of function which is corrected by ongoing treatment and conveys moderate risk (Table 1).
Following the loss of 2 large venous access sites suitable for parenteral nutrition and high flow infusions (i.e., internal jugular, subclavian, and femoral vein), each subsequent loss was given a score of 1.
3. Surgery
Patients were selected for transplantation according to the conventional indications as previously published [5, 6]. In summary, these were irreversible intestinal failure and inability to continue parenteral nutrition, due to such conditions as loss of venous access and parenteral nutrition-induced liver disease. Other indications included the need for extensive evisceration to remove desmoid tumors. Patients received an intestinal graft either in isolation or as part of a cluster of abdominal organs including liver (multivisceral) or excluding liver (modified multivisceral transplant). All patients received induction therapy with either Alemtuzumab (Campath 1H), Daclizumab (Zenapax), or a few with glucocorticosteroids. The surgical techniques and postoperative care were in keeping with those previously described [5].
4. Data Collection
The data were collected prospectively during routine preoperative evaluation at the transplant centres at the University of Cambridge, United Kingdom, and University of Miami, United States of America, from 1997 to 2003. The analysis was undertaken retrospectively. Patient data were consecutively entered into a database as a routine at each institution from which data were acquired for the analysis. Where comorbidity data were incomplete, patients were excluded from the analysis; this was undertaken without knowledge of their subsequent survival.
5. Statistics
The primary outcome measured was median recipient survival with 95% confidence interval (CI) and was calculated by Kaplan-Meier (KM) approach with comparison using logrank test. The accuracy of the comorbidity score to predict postoperative survival was tested by using Harrell’s C statistic which estimates the proportion of accurate predictions. A Harrell’s C index varies from 0.5 to 1 reflecting no discrimination to perfect discrimination, respectively.
Student’s -test for analysis of variance was used.
6. Results
The donor and patient demographics are detailed in (Tables 2 and 3, resp.). A total of 72 adults (M : F, 33 : 39) were included in the study.
Patients received either an isolated intestinal graft () or intestine in combination with other organs including liver (multivisceral) () or excluding liver (modified multivisceral) (). In 31 patients, the large intestine was also transplanted, and 6 received a kidney. Mean (SD) survival of the group was 1501 (1444) days. No repeat transplantations were included in the cohort.
The cause of intestinal failure in recipients was most commonly short gut syndrome (33%), and the most frequent procedure in this cohort was multivisceral transplantation (intestine, liver, and additional organs) in 38%, followed by isolated intestine (37%), and modified multivisceral (intestine and other organs excluding liver) 25% (Table 3).
Most patients received Alemtuzumab (Campath 1H) induction immunosuppression, and the mean (SD) number of rejection episodes per patient was (Table 4).
6.1. Kaplan-Meier Survival Analysis
Individual patient comorbidity scores and their postoperative survivals were found to correlate inversely ( −0.72; ). The Kaplan-Meier (KM) curve analysis of the cumulative recipient survival revealed a significant inverse association between patient survival and comorbidity score as evaluated by the logrank test for trend () (Figure 1). For the purposes of comparing survival chance, patients were grouped into comorbidity scores of 0 and 1, 2 and 3, 4 and 5, 6, and above. The hazard ratio (95% CI) for death (compared to group 0 + 1) was found to increase as the comorbidity score increased: 1.945 (0.7622–5.816), 5.075 (3.314–36.17), and 13.77 (463.3-120100), respectively; this became significantly greater than group 0 + 1 at group 4 + 5 () (Figure 1).
The survival of patients receiving isolated intestinal grafts was compared to those who underwent multivisceral transplantation. Kaplan-Meier analysis revealed no difference in survival between patients receiving isolated and multivisceral transplantation: logrank (Mantel-Cox) test: ; hazard ratio: 0.8129 (0.4599 to 1.366). Median survival for isolated transplants (2007 days) was similar to that for multivisceral grafts (3107 days).
6.2. Receiver-Operator Curve Analysis
The accuracy of the score to predict postoperative survival was evaluated using the receiver-operator characteristic analysis. A receiver-operator curve (ROC curve) was constructed for survival prediction at 1, 3, 5, and 10 years. The Harrell’s C statistic was calculated for each curve and represents the overall accuracy when assessed in terms of specificity and sensitivity at all possible points. The C statistic indicated very good or excellent accuracy of the comorbidity score at all times frames (Figure 2, Table 5).
7. Discussion
We recognise that the interpretation of these data should be undertaken in the context that there will be differences in the protocols of transplant centres which might have an influence on the relative importance of comorbidity on survival. For instance, the majority of patients in our cohort received Alemtuzumab as induction therapy which may have made the impact of comorbidity more significant. There will also be additional influences on survival, particularly in the long term, that are not included in this analysis. Nevertheless, in this cohort of patients, our results indicate that the degree of preoperative comorbidity appears to predict posttransplant survival following small intestinal and multivisceral transplantation in adults. The predictive value of the comorbidity score might allow better patient selection, focus on pretransplant optimization, and facilitate informed consent for the procedure. Patients who have a low comorbidity score might be considered for earlier transplantation or even preemptive transplantation, where transplantation is offered at the time of developing intestinal failure. This would be particularly appropriate for those with high risk circumstances, where they are likely to have a reduced life expectancy on PN [6], such as pseudoobstruction and systemic sclerosis and possibly ultrashort bowel syndrome (<30 cm of jejunum), rather than waiting for them to develop the traditional indications through complications of their PN. In better performing centres, it might be reasonable to offer patients preemptive transplantation on the basis of improving a very poor quality of life (QOL). Patients in comorbidity groups 0 + 1 have a 10-year survival similar or better than unselected patients on PN. However, before this becomes common practice, it will be important to develop methods of determining which patients will benefit in terms of QOL from transplantation, as it is clear that certain aspects of patient’s lives do not improve with transplantation [7]. Furthermore, the timing of transplantation might be influenced by the desire to limit comorbidity to keep the comorbidity score low. For instance, a patient with early comorbidity in several body systems might benefit from early transplantation before these deteriorate; conversely, patients with active comorbidity might have a better survival chance if transplantation is delayed until they receive treatment to stabilise concurrent illness.
As postoperative survival improves poor quality of life, it is likely to become a more frequent indication for transplantation. Under these circumstances, it will be crucial to predict postoperative survival for individual patients to better evaluate the risk/benefit ratio of transplantation and facilitate informed consent, and the comorbidity score might also be of considerable use for this purpose. As the role of other factors, such as the presence of HLA antibodies, becomes clearer, this scoring system may be improved by their inclusion [8].
In conclusion, the numeric score (comorbidity score) developed to allow us to semiquantify preoperative comorbidity appears to be an accurate and convenient method of assessing risk using preoperative risk factors. A future multicentre international prospective study to evaluate this survival risk factor is important to confirm our initial observation and may lead to a useful clinical tool in the selection and timing of patients for transplantation.
Abbreviations
| SD: | Standard deviation |
| CI: | Confidence interval |
| Pn: | Parenteral nutrition |
| DM: | Diabetes mellitus |
| COPD: | Chronic obstructive pulmonary disease |
| KM: | Kaplan-Meier |
| M: | Male |
| F: | Female |
| : | Spearman’s correlation coefficient |
| ROC: | Receiver-operator characteristic |
| QOL: | Quality of life |
| Yrs: | Years |
| g: | Grams |
| dL: | Decilitre |
| MVT: | Multivisceral. |
Conflict of Interests
The authors of this paper have no conflict of interests.
Authors’ Contribution
S. J. Middleton, R. Sivaprakasam, C. Pither, S. M. Gabe, M. Dawwas, N. V. Jamieson, and A. J. Butler took part in the design and analysis of the study and preparation of the paper. A. G. Tzakis, S. J. Middleton, E. R. Island, T. Hidenori, and S. Nishida took part in the methods and data collection.