2019 年 1 巻 2 号 p. 69-75
BACKGROUND
Previous studies have suggested that haptoglobin may be beneficial for preventing acute kidney injury (AKI) after severe burns. Although haptoglobin has been approved for the treatment of haemoglobinuria and subsequent AKI in Japan, robust evidence for this is lacking. We examined whether haptoglobin may be effective in preventing AKI requiring renal replacement therapy among patients with severe burns.
METHODS
We performed propensity score-matched analyses of the Japanese nationwide Diagnosis Procedure Combination inpatient database. We identified patients with severe burns (burn index ≥ 10) whose data were recorded from 1 July 2010 to 31 March 2013. We compared patients who were administered haptoglobin within 2 days of admission (haptoglobin group) and those who were not administered haptoglobin (control group). The main outcomes were: requirement for renal replacement therapy during admission, 28-day in-hospital mortality, length of hospital stay and ventilator-free days.
RESULTS
Eligible patients (n = 3223) from 618 hospitals were categorized into either the haptoglobin group (n = 263) or the control group (n = 2960). Propensity score matching created a matched cohort of 185 pairs with and without haptoglobin. There were no significant differences between the propensity score-matched groups in terms of the proportion of patients requiring renal replacement therapy (control vs haptoglobin, 25.9% vs 31.9%; p = 0.25), ventilator-free days (13 vs 6 days, p = 0.95), length of hospital stay (38 vs. 27, p = 0.45) and 28-day mortality (42.7% vs 46.5%, p = 0.53).
CONCLUSIONS
Our results suggest that haptoglobin use is not associated with reduced requirement for renal replacement therapy, increased ventilator-free days, length of hospital stay, or reduced 28-day mortality in patients with severe burns.
Acute kidney injury (AKI) is one of the major complications in patients with severe burns [1–5]. Burn patients with AKI are likely to have a longer stay in the intensive care unit and higher mortality [1–5]. Although the introduction of renal replacement therapy has reduced the mortality of patients with burns who have AKI, no effective pharmacologic treatment has been identified yet [6–9].
Emergence of free hemoglobin in the plasma, as a result of intravascular hemolysis after severe thermal injury, is one of the risk factors for acute renal dysfunction after thermal injury [6, 7]. After a thermal injury, free hemoglobin that is released from degraded red blood cells combines tightly with free haptoglobin and forms a haptoglobin–hemoglobin complex. This haptoglobin-hemoglobin complex allows free hemoglobin to bypass the kidney, which is carried to the liver and metabolized. However, if the production of free hemoglobin exceeds the plasma haptoglobin, unconjugated free hemoglobin passes through the kidney [6, 7, 10]. Thus, this relative haptoglobin deficiency may cause degenerative changes in tubular cells due to the formation of hemoglobin casts (with or without hemoglobinemia). Consequently, AKI may develop in this condition especially when combined with early burn shock [7].
Plasma-purified haptoglobin (Japan Blood Products Organization, Tokyo, Japan) has been approved for clinical use in Japan since 1985 for hemoglobinuria following profound hemolysis in severe burns, cardiovascular surgery with extracorporeal circulation usage, or massive transfusion provision [8, 11–15]. The rationale for the provision of haptoglobin is based on the concept that they act by scavenging circulating-free hemoglobin [10, 16–19]. However, haptoglobin provision in patients with burns has been described only in a few case reports [14, 20, 21] and one small case-series (n = 10) from Japan [11]. Recently, Depret et al. [22] reported in their single-center retrospective cohort study (n = 130) that undetectable (low plasma) haptoglobin was associated with major adverse kidney events in critically ill patients with burns. These data suggested that provision of haptoglobin may be rational for preventing AKI in severe burns; however, no robust clinical evidence supports this idea.
We hypothesized that haptoglobin administration could be effective in preventing AKI requiring renal replacement therapy among patients with severe burns. We aimed to evaluate our hypothesis using data from a large Japanese nationwide inpatient database, the only country where haptoglobin is approved for use in clinical settings. We also evaluated the relationships between provision of haptoglobin and 28-day in-hospital mortality, length of hospital stay, and ventilator-free days in this study.
This study was approved by the Institutional Review Board of the University of Tokyo, which waived the requirement for informed patient consent because of the anonymous nature of the data.
DATA SOURCE AND VARIABLESWe extracted and analyzed the data from the Japanese nationwide Diagnosis Procedure Combination (DPC) database, the details of which have been described previously, elsewhere [23–25]. In short, the DPC database includes summary and administrative claims data for all patients discharged from more than 1000 hospitals in Japan. It covers over 90% of all tertiary-care emergency hospitals and more than 90% of burn-specialists training centers certified by the Japanese Society for Burn Injuries [26–28]. The DPC database includes each person’s gender, age, status and discharge date in previous admission, primary diagnosis, comorbidities (on admission), and post-admission complications. The diseases are all coded by the International Classification of Diseases, 10th Revision codes (ICD-10). Japan Coma Scale (JCS) scores on admission, all medical and surgical procedures, records of all drug and devices provided, and discharged day with status (i.e. either to home, to another hospital/institution, or as a result of death) were also recorded [26–28]. Although the percentage of total burn surface area was not recorded in the database, the burn index [28, 29] was available. The burn index considers both the area and thickness of the burn area (i.e. burn index = full thickness of total burn surface area + 1/2 partial thickness of total burn surface area) [28, 29]. Thus, the burn index is thought to be a better predictor of mortality and morbidity compared to simple percentage of total body surface area [30]. To guarantee the validity of the coding, the physicians in charge have an obligation to record the diagnoses with reference to the medical charts [26–28].
We categorized the JCS scores into four groups: 0 (alert); 1–3 (delirium); 10–30 (somnolence); and 100–300 (coma) [26, 27]. Charlson Comorbidity Index (CCI) was calculated from the ICD-10 code for each comorbidity [31]. Hospital volume was categorized into two (low and high), as defined and the number of eligible patients with burns admitted in the participating hospitals.
PATIENT SELECTION AND ENDPOINTSWe compared patients with severe burns who were administered haptoglobin within 2 days after admission (haptoglobin group) and those who were not administered haptoglobin (control group) in the DPC database from 1 July 2010 to 31 March 2013. Severe burns were defined as burn index ≥10 [26–28]. The exclusion criteria for the current study were: 1) out-of-hospital cardiac arrest, 2) discharged within 2 days after admission (i.e. in order to avoid immortal time bias) [32], and 3) readmitted patients with burns.
The primary endpoint of the study was requirement of renal replacement therapy (continuous hemodiafiltration and/or hemodialysis) during admission. Secondary outcomes included 28-day in-hospital mortality, length of hospital stay, and ventilator-free days (VFDs) (i.e. defined as the number of days the patient remained alive without mechanical ventilation assistance during the first 28 days after admission; patients who died before day 28 were assigned with 0 day) [33].
STATISTICAL ANALYSISWe performed a one-to-one propensity score matching analysis between the haptoglobin and control groups [34, 35]. We estimated the propensity score by fitting a logistic regression model for haptoglobin use as a function of the patients’ clinical characteristics, hospital factors, and demographics, which were reported in the previous literature to have the potential to affect kidney dysfunction and mortality in severe burns: age; gender; hospital volume; CCI, burn index; JCS; use of mechanical ventilation within two days of admission, catecholamines (dopamine, dobutamine, and/or norepinephrine), antibiotics, red blood cell, fresh frozen plasma, platelet, albumin, antithrombin, heparin, ulinastatin, thrombomodulin, and nafamostat mesylate; and requirement for surgical procedure (debridement and escharotomy) [5–8, 12, 26–29, 36–40]. We calculated the C-statistics for evaluating the goodness-of-fit. By using nearest-neighbor matching, we performed one-to-one propensity score matching (i.e. a match occurs when a haptoglobin group patient had an estimated score within 0.2 standard deviations of the control group patient [34]). Standardized differences were used to examine the balance of baseline variables, and a value less than 10% was regarded as well-balanced [34]. Data are presented as numbers with percentages or median with interquartile range (25%–75% value). Mann-Whitney U test were used for evaluating continuous variables and χ2 tests or Fisher’s exact tests were used for categorical variables when comparing two groups. We considered a value of p < 0.05 as statistically significant. Given the study’s retrospective nature, we did not perform sample-size estimation for the present study. IBM SPSS version 22 (IBM Corp., Armonk, NY, USA) was used for all statistical analyses.
From the 618 hospitals during the 33-month study period, a total of 3223 patients were identified as eligible for the present study. Patients were divided into the control and haptoglobin groups (control, n = 2960; haptoglobin, n = 263). One hundred and eighty-five propensity score-matched pairs (n = 370) were extracted (Fig. 1). The C-statistic was 0.93 for the modelling of propensity score.
Table 1 represents the baseline patients’ characteristics among the unmatched and propensity score-matched groups. Among the unmatched groups, patients were more likely to receive haptoglobin when they had a higher burn index, and more catecholamines, mechanical ventilation, and other treatments were generally required for severe burns. On the other hand, the baseline characteristics were well balanced among the propensity score-matched groups.
| Unmatched groups | Matched groups | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Variable | Control (n = 2960) |
Haptoglobin (n = 263) |
Standardized differences, % |
Control (n = 185) |
Haptoglobin (n = 185) |
Standardized differences, % |
||||
| Age, years (SD) | 55.9 | (26.7) | 62.8 | (20.0) | −29.2 | 66.1 | (17.5) | 65.9 | (16.2) | 1.2 |
| Adult | 2604 | (88.0) | 258 | (98.1) | −40.6 | 181 | (97.8) | 183 | (98.9) | −8.6 |
| Sex (male) | 1813 | (61.3) | 171 | (65.0) | −7.8 | 116 | (62.7) | 123 | (66.5) | −7.9 |
| Burn index, mean (SD) | 20.5 | (15.7) | 45.7 | (23.7) | −125.1 | 37.4 | (23.8) | 39.0 | (21.1) | −7.1 |
| Charlson comorbidity index ≥ 1 | 401 | (13.5) | 19 | (7.2) | 20.8 | 21 | (11.4) | 17 | (9.2) | 7.1 |
| High volume hospital, cases | 918 | (31.0) | 127 | (48.3) | −35.9 | 90 | (48.6) | 83 | (44.9) | 7.6 |
| Consciousness level | ||||||||||
| Alert | 2203 | (74.4) | 89 | (33.8) | 89.2 | 68 | (36.8) | 65 | (35.1) | 3.4 |
| Delirium | 431 | (14.6) | 66 | (25.1) | −26.7 | 57 | (30.8) | 51 | (27.6) | 7.1 |
| Somnolence | 120 | (4.1) | 24 | (9.1) | −20.5 | 13 | (7.0) | 15 | (8.1) | −4.1 |
| Coma | 206 | (7.0) | 84 | (31.9) | −66.5 | 47 | (25.4) | 54 | (29.2) | −8.5 |
| Mechanical ventilation | 553 | (18.7) | 222 | (84.4) | −174.6 | 156 | (84.3) | 149 | (80.5) | 9.9 |
| Antibiotics use | 1164 | (39.3) | 179 | (68.1) | −60.2 | 117 | (63.2) | 120 | (64.9) | −3.4 |
| Dopamine use | 190 | (6.4) | 87 | (33.1) | −71.1 | 58 | (31.4) | 55 | (29.7) | 3.5 |
| Dobutamine use | 40 | (1.4) | 20 | (7.6) | −30.6 | 10 | (5.4) | 12 | (6.5) | −4.6 |
| Noradrenaline use | 117 | (4.0) | 50 | (19.0) | −48.6 | 33 | (17.8) | 34 | (18.4) | −1.4 |
| Red blood cell use | 93 | (3.1) | 55 | (20.9) | −56.8 | 27 | (14.6) | 31 | (16.8) | −5.9 |
| Fresh frozen plasma use | 194 | (6.6) | 104 | (39.5) | −85.1 | 54 | (29.2) | 59 | (31.9) | −5.9 |
| Platelet use | 19 | (0.6) | 15 | (5.7) | −29.2 | 7 | (3.8) | 7 | (3.8) | 0.0 |
| Albumine use | 541 | (18.3) | 200 | (76.0) | −141.9 | 130 | (70.3) | 129 | (69.7) | 1.2 |
| Antithrombin use | 85 | (2.9) | 67 | (25.5) | −68.5 | 24 | (13.0) | 29 | (15.7) | −7.7 |
| Hepaline use | 565 | (19.1) | 189 | (71.9) | −125.0 | 125 | (67.6) | 121 | (65.4) | 4.6 |
| Thorombomoduline use | 30 | (1.0) | 14 | (5.3) | −24.8 | 6 | (3.2) | 9 | (4.9) | −8.2 |
| Nafamostat use | 53 | (1.8) | 23 | (8.7) | −31.5 | 15 | (8.1) | 16 | (8.6) | −2.0 |
| Ulinastatine use | 40 | (1.4) | 30 | (11.4) | −42.0 | 18 | (9.7) | 14 | (7.6) | 7.7 |
| Debridement performed | 119 | (4.0) | 18 | (6.8) | −12.5 | 25 | (13.5) | 28 | (15.1) | −4.6 |
| Escharotomy performed | 89 | (3.0) | 64 | (24.3) | −65.3 | 14 | (7.6) | 13 | (7.0) | 2.1 |
SD, standard deviation
As shown in Table 2, there were no significant differences between the control and haptoglobin groups of the propensity score-matched groups in terms of the proportion of patients requiring renal replacement therapy (25.9% vs 31.9%; p = 0.25), VFDs (13 vs 6 days, p = 0.95), length of hospital stay (38 vs. 27 days, p = 0.45) and 28-day mortality (42.7% vs 46.5%, p = 0.53).
| Control (n = 185) | Haptoglobin (n = 185) | P value | |||
|---|---|---|---|---|---|
| Renal replacement therapy, n (%) | 48 | (25.9) | 59 | (31.9) | 0.25 |
| Ventilator-free days, median (25%–75%) | 13 | (0–28) | 6 | (0–28) | 0.95 |
| Length of hospital stay, median (25%–75%) | 38 | (8–82) | 27 | (4–95) | 0.45 |
| Morality 28-day, n (%) | 79 | (42.7) | 86 | (46.5) | 0.53 |
In this study, we extracted and analyzed data from a Japanese nationwide database of severe burns. We could not find a significant association between haptoglobin provision and requirement of renal replacement therapy, better VFDs, length of hospital stay and 28-day mortality.
Previously, several experimental animal studies suggested that haptoglobin provision may be useful for the prevention of an AKI after severe burns [10, 16, 18–20]. However to date, only a few clinical case reports and a few small clinical studies support the idea of haptoglobin provision for kidney protection after a thermal injury [11, 12, 14, 20]. Currently, haptoglobin is approved to be used in clinical practice in Japan. Moreover, the treatment guidelines of the Japanese Society for Burn Injuries [8] suggest the use of haptoglobin in case of hemoglobinuria after thermal injury (Recommendation grade B). However, no other international treatment guidelines mention haptoglobin treatment for AKI after burns because of a lack of scientific evidence [9].
Recently, Kubota et al. [15] reported in their single-center retrospective study that haptoglobin provision during cardiovascular surgery was associated with a lower risk of post-operative AKI. More recently, Depret et al. [22] reported that undetectable plasma haptoglobin was associated with major adverse kidney events in critically ill patients with burns. They provided a possible rationale for a biomarker-guided therapy using haptoglobin in patients with severe burns [22].
This study is the first large-scale clinical data on haptoglobin provision and AKI after severe burns. The methodological strengths of the present study include its use of a large-scale database and focus onto the patients with severe burns only (i.e. burn index > 10 [28]). Most renal failures occur either immediately (i.e. within 3 days) after the injury or at a later period when sepsis develops [6, 7]. Late-onset renal failure is usually the consequence of sepsis and is often associated with other organ failure [6, 7]. Thus, the latter one is not our target patients at least for the current study. On the other hand, for prevention for developing early renal failure, it may be reasonable to start haptoglobin before 3 days. As per our several previous burn studies (e.g. prophylactic antibiotic for burns [26] and antithrombin use for burns [27]) we compared patients with severe burns who were administered haptoglobin within 2 days after admission and those who were not administered haptoglobin for the current study. The baseline patient characteristics in the unmatched groups suggested that haptoglobin tended to be used in more critical patients with severe burns. However, we ensured the comparability of the two groups by using propensity score matching, which provides a sound method for constructing a randomized experiment-like situation by comparing groups with similar observed characteristics, without specifying the relationships between confounders and outcomes [34]. When estimating propensity score, we were able to include all the major factors (for example, patient’s age, gender, size and depth of the burn, and the use of mechanical ventilation) that could potentially affect the incidence of AKI and mortality of patients with severe burns [5, 7, 28, 29, 37, 38, 41]. We calculated the C-statistics of 0.93 for evaluating the goodness-of-fit for the propensity score. The high area under the curve suggests that the model greatly depended on the variables we selected. After the one-to-one propensity score matching, all the measured variables were well balanced between the haptoglobin and control groups. The results of our current study suggested that patients with severe burns who received haptoglobin were not significantly associated with better outcomes in terms of renal replacement therapy, VFDs, and 28-day mortality. Although we should wait for the evidence from future interventional studies to confirm our results, we speculate that there may be a lack of benefit for haptoglobin administration for early phase of AKI after severe burns. Haptoglobin is human blood products donated from volunteers. It costs approximately ¥87,138 (i.e. approximately US$ 810 or 660 EUR) for one administration. Thus, we suggest that haptoglobin should be administered with caution for patients with severe burns unless new evidence appears.
The current study has several limitations. First, the present study was retrospective in nature. Even though the propensity score matching method was used to adjust for differences in baseline characteristics and disease severity (i.e. burn index), unmeasured confounding factors might have existed and led to a bias. Unreported possible confounding parameters, not available from the DPC database, include baseline levels of haptoglobin, kidney function, fluid balance including urinary output per day, delays in care, and other vital signs. Second, we focused on the patients with severe burns in the current study. Thus, we cannot generalize the current results to treatments for other circumstances such as massive transfusion and cardiac surgery cases. Third, we could not exclude patients receiving chronic dialysis because of lack of data. Some patients with chronic dialysis in the control group (but possibly none in the haptoglobin group) might have been receiving long-term dialysis, which may have worsened the outcomes of the group. Unfortunately, we could not determine such patients from our dataset.
There is no doubt that randomized control trials are the gold standard to evaluate the effectiveness of particular interventions. We cannot draw any robust conclusions regarding the efficacy of haptoglobin in severe burn patients in general yet, at least from our retrospective analysis. In studies assessing haptoglobin administration, it is necessary to include plasma haptoglobin levels among the inclusion criteria for future studies.
Analysis of the Japanese nationwide database suggests that haptoglobin use may not be associated with a reduced requirement for renal replacement therapy, increased VFDs, length of hospital stay and reduced 28-day mortality in patients with severe burns. Further studies are required to confirm our results.
This work was supported by Grants from the Ministry of Health, Labour and Welfare, Japan (Grant Numbers: H29-Policy-Designated-009 to Prof. Fushimi and Prof. Yasunaga; H29-ICT-General-004 to Prof. Yasunaga), Grants-in-Aid for Scientific Research, Japan (Grant Number: KAKENHI-15H05685 and KAKENHI-16KK0211 to Dr. Tagami, and KAKENHI-17H04141 to Dr. Yasunaga), and a Grant from the Japan Agency for Medical Research and Development (AMED). The funders had no role in the execution of this study or interpretation of the results.
The authors declare that they have no competing interests.
