Characteristics of Acute Kidney Injury in Patients Infected with the 2009 Influenza A (H1N1) Virus

Abstract

Background and objectives: There have been few studies investigating acute kidney injury (AKI) in patients infected with the 2009 pandemic influenza A (H1N1) virus. Therefore, the objective of this study was to identify the factors associated with AKI in H1N1-infected patients.

Design, setting, participants, & measurements: This was a study of 47 consecutive critically ill adult patients with reverse transcriptase-PCR-confirmed H1N1 infection in Brazil. Outcome measures were AKI (as defined by the Risk, Injury, Failure, Loss, and End-stage renal failure [RIFLE] criteria) and in-hospital death.

Results: AKI was identified in 25 (53%) of the 47 H1N1-infected patients. AKI was associated with vasopressor use, mechanical ventilation, high Acute Physiology and Chronic Health Evaluation II (APACHE II) scores, and severe acidosis as well as with higher levels of C-reactive protein and lactic dehydrogenase upon intensive care unit (ICU) admission. A nephrology consultation was requested for 16 patients (64%), and 8 (50%) required dialysis. At ICU admission, 7 (15%) of the 25 AKI patients had not yet progressed to AKI. However, by 72 hours after ICU admission, no difference in RIFLE score was found between AKI survivors and nonsurvivors. Of the 47 patients, 9 (19%) died, all with AKI. Mortality was associated with mechanical ventilation, vasopressor use, dialysis, high APACHE II score, high bilirubin levels, and a low RIFLE score at ICU admission.

Conclusions: Among critically ill H1N1-infected patients, the incidence of AKI is high. In such patients, AKI is mainly attributable to shock.

To date, there have been few studies of renal alterations in patients infected with the influenza A (H1N1) virus. A study conducted in Mexico identified acute kidney injury (AKI) in 6 of 18 patients hospitalized for H1N1 infection, and 5 of those 6 patients died (1). In an autopsy study involving fatal cases of H1N1 infection in Brazil, all 21 of the individuals autopsied were found to have mild/moderate acute tubular necrosis, myoglobin being identified in the tubules of 4, and 1 being diagnosed with thrombotic microangiopathy (2). However, those findings were attributed to multiple organ failure and AKI was identified in only nine patients. The World Health Organization (WHO) reported that rhabdomyolysis with renal failure and myocarditis can be present in H1N1 infection (3), and that approximately 50% of the H1N1-infected patients hospitalized in Mexico had some degree of renal dysfunction caused by rhabdomyolysis or hypotension (4). Nevertheless, AKI in H1N1-infected patients has yet to be well characterized. Here, we present the clinical and biochemical characteristics of AKI in critically ill H1N1-infected patients, including the details of dialysis therapy and outcomes.

Materials and Methods

The Hospital das Clínicas is a university hospital located in the largest city in Brazil, São Paulo, where the H1N1 epidemic peaked during July and August 2009. All critically ill adult patients with suspected or confirmed H1N1 infection were initially admitted to the intensive care unit (ICU) of the Department of Infectious and Parasitic Diseases, where 52 cases of confirmed H1N1 infection were treated between July 1 and August 31. The diagnosis of H1N1 infection was confirmed by reverse transcriptase PCR (RT-PCR) of nasopharyngeal swab samples. In two patients, we also performed RT-PCR in urine samples, and the results were positive in one. We defined AKI using the Risk, Injury, Failure, Loss, and End-stage renal failure (RIFLE) criteria solely on the basis of an increase in serum creatinine (S_creat) from baseline to 72 hours after ICU admission (5). Only the presence or absence of oliguria was documented. Five patients were excluded: three because S_creat was measured only at ICU admission and two because AKI developed after 72 hours and was due to secondary infection. Therefore, the final sample comprised 47 patients. A baseline GFR <60 ml/min per 1.73 m² was considered indicative of preexisting renal dysfunction. We estimated GFR using the abbreviated Modification of Diet in Renal Disease equation (6). Transplant recipients, patients undergoing chemotherapy, and patients receiving immunosuppressive therapy were classified as immunosuppressed. There were two primary outcome measures: AKI within the first 72 hours after ICU admission and in-hospital death. We also evaluated S_creat at 3 months after hospital discharge. There was no active surveillance for AKI cases (in the ICU of the Hospital das Clínicas, nephrology consultations are requested by the attending physicians). We evaluated two groups HIN1-infected patients: those who developed AKI and those who did not.

We evaluated S_creat at baseline (last measurement before the onset of influenza symptoms) and at ICU admission and registered the highest S_creat level within the first 72 hours after ICU admission and the lowest S_creat level within the first 3 months after hospital discharge. For patients who required dialysis, we also registered the S_creat level before the first dialysis. Other biochemical parameters were evaluated only at ICU admission. Complete recovery of renal function was defined as a postdischarge S_creat level ≤10% higher than that registered at baseline.

The Hospital das Clínicas Research Ethics Committee approved the study protocol (no. 0958/09). In view of the observational (noninterventional), retrospective nature of the study, with guaranteed confidentiality, the need for informed consent was waived.

Statistical Analyses

We used the Kolmogorov–Smirnov test to evaluate the Gaussian distribution of continuous variables, which were compared using t test or the Mann–Whitney test, as appropriate, and are expressed as mean ± SD or median (interquartile range), respectively. We used Fisher's exact test or the χ² test to compare categorical variables, which are expressed as absolute values and percentages. All analyses were performed using GraphPad Prism version 4.00 (GraphPad Software, San Diego, CA). Values of P < 0.05 were considered significant.

Results

Our sample comprised 25 women (2 of whom were pregnant) and 22 men. The mean age was 43 ± 15 years. A history of comorbidities was observed in all but six patients: immunosuppression (n = 10), chronic pulmonary disease (n = 10), diabetes (n = 7), hypertension (n = 7), neurologic disease (n = 5), and cardiopathy (n = 4). Although nine patients were obese, none were morbidly obese. Five patients (11%) had a baseline GFR <60 ml/min per 1.73 m². All patients were treated with oseltamivir at 75 to 150 mg twice a day, with the dosage reduced to 75 mg once daily in patients with an estimated GFR <30 ml/min. An additional 75 mg were administered after each dialysis session. Only five patients received oseltamivir within the first 48 hours after the onset of influenza symptoms.

AKI Characteristics

A diagnosis of AKI was made in 25 (53%) of the 47 patients evaluated. Stays in the ICU were longer for patients with AKI than for those without (11 [6 to 24.5] versus 5 [2 to 8] days, P = 0.005). The clinical and laboratory characteristics of the AKI and non-AKI groups are shown in Tables 1 and 2. All but three of the AKI group patients had a history of comorbidities. Urinalysis was performed in 15 (60%) of the 25 AKI group patients and in 10 (45%) of the 22 non-AKI group patients. The two groups were similar in terms of the urinary parameters, except for proteinuria, which was more common in the AKI group (100% versus 50%, P = 0.004), in which it was also more pronounced. Hematuria was found in six (40%) of the AKI group patients and in two (20%) of the non-AKI group patients (P = 0.401). Granular casts were present in eight (53%) of the AKI group patients and in four (40%) of the non-AKI group patients (P = 0.688). Urate crystals were present in six (40%) of the AKI group patients and in two (20%) of the non-AKI group patients (P = 0.401). Urine samples of two AKI patients were tested for H1N1, and the result was positive in one. Applying the RIFLE criteria, we found that at ICU admission, 11 patients were classified as not having AKI, 7 were classified as at risk for kidney injury, 2 were classified as having kidney injury, and 5 were classified as having renal failure. At 72 hours after ICU admission, 11 patients were classified as at risk for kidney injury, 4 were classified as having kidney injury, and 10 were classified as having renal failure. A nephrology consultation was requested for 16 (64%) of the 25 AKI patients.

Table 1.

Characteristics of H1N1-infected patients with and without AKI

Characteristic	With AKI (n = 25)	Without AKI (n = 22)	P
Age (years), mean ± SD	44 ± 17	42 ± 12	0.59
Gender, male/female	12/13	10/12	1.00
Oseltamivir ≥48 hours after symptom onset, yes/no	20/5	18/4	1.00
Days from hospitalization to ICU admission, median (range)	1 (0.5 to 2.5)	1 (0 to 2)	0.33
Comorbidities
previous renal dysfunction, yes/no	2/23	3/19	0.65
chronic respiratory disease, yes/no	6/19	5/17	1.00
immunosuppression, yes/no	4/21	8/14	0.18
hypertension, yes/no	5/20	1/21	0.19
diabetes, yes/no	4/21	1/21	0.35
obesity,c yes/no	5/20	4/18	1.00
pregnancy, yes/no	2/11	0/12	0.48
At ICU admission
mechanical ventilation, yes/no	19/6	12/10	0.01
vasopressor use, yes/no	19/6	3/19	<0.001
APACHE II score, mean ± SD	17 ± 5a	12 ± 7b	0.008
PaO2/FiO2 ≤ 200, yes/no	18/6	3/15	<0.001
Evolution
length of hospital stay (days), median (range)	17 (8.5 to 33)	6 (3 to 15)	0.001
death, yes/no	9/16	0/22	0.001

PaO₂, arterial oxygen tension; FiO₂, fraction of inspired oxygen.

^an = 24,

^bn = 20,

^cNo morbid obesity.

Table 2.

Laboratory results, at ICU admission, for patients with and without AKI

Variable	With AKI (n = 25)	Without AKI (n = 22)	P
Screat (mg/dl), median (range)	1.04 (0.70 to 1.75)	0.95 (0.66 to 1.14)	0.32
pH, mean ± SD	7.27 ± 0.13a	7.36 ± 0.07b	0.016
Bicarbonate (mEq/L), mean ± SD	20.86 ± p 5.87a	19.56 ± 5.24b	0.45
Albumin (g/dl), mean ± SD	2.80 ± 0.58b	3.22 ± 0.64c	0.05
CK (IU/L), median (range)	197 (102.5 to 649.5)	166.5 (60 to 717.5)	0.54
LDH (IU/L), median (range)	1001 (519 to 1934)a	546.5 (410.5 to 890.5)	0.02
CRP (mg/dl), median (range)	155 (102 to 229)	104.5 (26.7 to 161)	0.02
Lactate (mg/dl), median (range)	25 (15.5 to 35.5)a	21 (14 to 27)b	0.27
Bilirubin (mg/dl), median (range)	0.6 (0.31 to 0.93)d	0.38 (0.26 to 0.62)e	0.10

CK, creatine kinase; LDH, lactate dehydrogenase; CRP, C-reactive protein (reference: <3 mg/dl).

^an = 24,

^bn = 19,

^cn = 14,

^dn = 23,

^en = 21.

Dialysis

Nine (36%) of the 25 AKI patients required dialysis, corresponding to 56% of the 16 H1N1-infected patients followed by the Nephrology Division staff. Of those nine patients, one had a venous catheter inserted but died before starting the procedure, and another (because of a lack of beds in the ICU) was transferred to another hospital, underwent dialysis, and eventually died. A senior nephrologist determined the indication for and type of dialysis on the basis of the clinical profile of each patient. Hypervolemia was an indication for dialysis in eight of the nine dialyzed patients, and five of the nine had presented with oliguria at ICU admission. Details of the dialysis procedures are presented in Table 3.

Table 3.

Characteristics of the initial dialysis, total time on dialysis, and changes of dialysis method

Patient	Screat (mg/dl)	Indication	Method	Anticoagulationa	Vascular Accessb	Duration (hours)	Volume Removed (L/session)	Time on Dialysis (days)	Change of Method
1	1.81	Acidosis + hypervolemia	CVVH	No	Subclavian	168	20.2	11	Yesc
2	5.57	Uremia	CVVH	Citrate	Femoral	14	0	23	Yesc
3	1.7	Acidosis + hypervolemia	CVVH	Heparin	Femoral	12	0.7	3	No
4d	1.62	Acidosis + hypervolemia	CVVH		Femoral
6	4.34	Uremia + hypervolemia	CVVH	Citrate	Femoral	7	0.7	22	Yesc
7e	1.9	Acidosis + hypervolemia	CVVH	NA	NA	NA	NA	NA	NA
13	4.88	Uremia + hypervolemia	CVVH	Citrate	Femoral	15	1.5	17	Yesc
15	2.53	Uremia + hypervolemia	SLED	No	Femoral	8	2.5	3	Yesf
16	1.02	Hypervolemia	SLED	No	Femoral	5	1.5	4	No

CVVH, continuous venovenous hemofiltration; NA, not available; SLED, sustained low-efficiency dialysis.

^aPatients with pulmonary bleeding did not receive anticoagulation.

^bTemporary catheters were employed in all patients.

^cChanged to SLED during the ICU stay.

^dPatient had a catheter inserted but died before starting dialysis.

^ePatient was transferred to and underwent CVVH at another hospital.

^fChanged to intermittent hemodialysis after ICU discharge.

Mortality

Of the 25 AKI patients, 9 (36%) died, accounting for all of the in-hospital deaths occurring in the study sample as a whole. Among the patients who developed AKI, in-hospital death was found to be associated with vasopressor use, mechanical ventilation, dialysis, and high levels of bilirubin as well as with not having AKI at ICU admission (Tables 4 and 5).

Table 4.

Clinical data related to nonsurviving and surviving acute kidney injury patients

Variable	Nonsurvivors (n = 9)	Survivors (n = 16)	P
Age (years), mean ± SD	48 ± 17	42 ± 17	0.41
Gender (male/female)	5/4	7/9	0.68
Oseltamivir ≥48 hours after symptom onset, yes/no	7/2	13/3	1.00
Days from hospitalization to ICU admission, median (range)	3 (0.5 to 1.5)	1 (0.5 to 1.5)	0.08
Comorbidities
previous renal dysfunction, yes/no	0/9	2/14	0.52
chronic respiratory disease, yes/no	2/7	4/12	1.00
immunosuppression, yes/no	2/7	2/14	0.60
hypertension, yes/no	1/8	5/11	0.36
diabetes, yes/no	1/8	5/11	0.36
obesity,b yes/no	1/8	5/11	0.36
pregnancy, yes/no	1/3	1/8	1.00
At ICU admission
AKI, yes/no	2/7	12/4	0.01
oliguria,c yes/no	2/5	3/4	1.00
mechanical ventilation, yes/no	9/0	10/6	0.004
vasopressor use, yes/no	7/2	3/13	0.009
APACHE II score, mean ± SD	21 ± 7.2a	15 ± 3.0	0.018
PaO2/FiO2 ≤ 200,c yes/no	9/0	10/5	0.11
Evolution
dialysis, yes/no	6/3	3/13	0.030
length of hospital stay (days), median (range)	28 (IQR 13 to 43)	14 (7.5 to 35)	0.30

IQR, interquartile range.

^an = 8.

^bNo morbid obesity.

^cData not reported for all patients.

Table 5.

Laboratory results at ICU admission for nonsurvivors and survivors of AKI

Variable	Nonsurvivors (n = 9)	Survivors (n = 16)	P
Screat (mg/dl), median (range)	0.60 (0.45 to 1.23)	1.17 (0.9 to 2.26)	0.023
pH, mean ± SD	7.30 ± 0.11	7.26 ± 0.14a	0.53
Bicarbonate (mEq/L), median (range)	20.2 (15 to 25.4)	20.3 (17.9 to 22.6)	0.59
Albumin (g/dl), mean ± SD	2.53 ± 0.51b	2.95 ± 0.57c	0.12
CK (IU/L), median (range)	248 (74.5 to 675.5)	188 (105.5 to 649.5)	0.97
LDH (IU/L), median (range)	891.5 (580 to 1543)	1170 (483 to 2143)	0.97
CRP (mg/dl), median (range)	157 (110.5 to 316)	140.5 (99.8 to 177.5)	0.41
Lactate (mg/dl), mean ± SD	29.5 ± 15.7	26.6 ± 12.6a	0.62
Bilirubin (mg/dl), median (range)	0.92 (0.76 to 1.00)d	0.47 (0.30 to 0.62)a	0.04

^an = 15,

^bn = 7,

^cn = 12,

^dn = 8.

Renal Function Recovery

None of the patients remained on dialysis after hospital discharge. After hospital discharge, S_creat was evaluated in 13 (81%) of the 16 surviving AKI patients, and the median value was 0.89 (0.65 to 1.23). By 3 months after hospital discharge, eight patients achieved complete recovery of renal function and five achieved partial recovery. The two patients with a history of renal dysfunction achieved complete recovery.

Discussion

There is a paucity of data regarding AKI in H1N1 infection. In a study involving 426 H1N1-infected patients in China, the authors reported the mean S_creat level (67 ± 20 μmol/L) but did not mention AKI (7). Data from Chile show that, from May to September 2009, 1562 patients were hospitalized with confirmed H1N1 infection, and 25% of those patients presented elevated S_creat levels (8). More recently, two studies analyzing AKI in critically ill patients in Canada and Argentina showed the incidence of AKI to be 66.7% and 63.6%, respectively (9,10), compared with the 53% observed in the study presented here. This difference might be attributable to the strict criteria we applied to select a sample composed only of patients with AKI due to H1N1 infection: we excluded three patients for whom S_creat had been determined only once, thereby making it impossible to confirm AKI, as well as two patients who developed AKI due to a secondary infection occurring more than 72 hours after ICU admission.

The RIFLE criteria, which were used to define AKI in the study conducted in Canada and in our study, constitute a highly sensitive diagnostic tool (5). The authors of the study conducted in Canada found the incidence of AKI to be 66.7% during the entire hospital stay and 47.9% on day 1 in the ICU (9). In our study, only 14.9% of the patients had AKI at ICU admission, but 29.8% developed AKI within the next 72 hours. This difference is likely related to the fact that our patients, in comparison with those evaluated in the other study, presented with disease that was less severe, as evidenced by the respective mean APACHE II scores (14.8 ± 6.5 versus 19.2 ± 6.5).

In our study, mechanical ventilation was strongly associated with the development of AKI: 76% of the AKI group patients required mechanical ventilation, compared with 54% of those in the non-AKI group (P = 0.01). Other factors associated with AKI were an arterial oxygen tension/fraction of inspired oxygen ratio <200, vasopressor use, high APACHE II scores, elevated C-reactive protein levels, elevated lactate dehydrogenase levels, and low pH. A history of comorbidities was not associated with AKI. In the study conducted in Canada, the factors associated with AKI were older age, higher body mass index, presence of asthma, and higher APACHE II scores (9). In the study conducted in Argentina, 95.5% of the patients required mechanical ventilation and the following factors were associated with AKI: pregnancy; immunosuppression; high APACHE II, Sequential Organ Failure Assessment, and Murray scores; hemodynamic instability; less time on mechanical ventilation; and thrombocytopenia (10). The discrepancies between our study and others can be attributed to differences in study design or in the populations studied (e.g., in the study conducted in Canada, 72% were female and 48% were Aboriginal).

In the study presented here, AKI was just one aspect of H1N1-related illness that was more severe, as indicated by the higher APACHE II scores and greater functional impairment (pulmonary and hepatic) among the AKI patients. Concomitant H1N1 infection and AKI was strongly associated with the development of hemodynamic instability during the ICU stay: 79% of our H1N1-infected AKI patients required vasopressor support. Although rhabdomyolysis has been implicated in the development of AKI (3), rhabdomyolysis, as evaluated by creatine kinase levels, was not found to be associated with AKI in our study or in the studies conducted in Canada and Argentina (9,10). It is noteworthy that 100% of our AKI patients had proteinuria. The fact that the H1N1 virus was detected in the urine of a patient tested early might indicate an early direct renal effect of the virus. Fislová et al. studied the pathogenetic profiles of three strains of human influenza A virus, including the A/PR/8/34 (H1N1) strain, which is highly virulent for mice (11). The mice were infected intranasally and monitored through RT-PCR detection of the virus in respiratory and nonrespiratory tissues. The authors identified viral RNA in the lung, heart, thymus, liver, spleen, and kidneys. Li et al. compared cell lines infected with pandemic H1N1, seasonal H1N1, and avian influenza A (H5N1) in terms of their growth characteristics (12). They found that the H1N1 viruses were able to replicate in hepatic, renal, neural, and monocyte cell lines. In a study conducted by To et al. (13), the viral loads of 22 pandemic H1N1 patients and 44 seasonal influenza historical controls were studied. The authors detected the H1N1 virus in the stool and urine samples of 4 (of 9) and 1 (of 14) patients, respectively. Taken together, these findings suggest a potential effect of the H1N1 virus on human kidneys.

There have been no studies evaluating renal function in H1N1-infected AKI patients after hospital discharge. In the study conducted in Canada, 11% of the survivors remained on dialysis after hospital discharge (9). In the study presented here, none of the AKI survivors were dialysis dependent at hospital discharge, and by 3 months after hospital discharge, complete recovery of renal function was achieved in 61% of the AKI survivors. We found no association between partial recovery of renal function and previous renal dysfunction. All of our AKI patients had proteinuria at ICU admission, and 39% later achieved partial recovery of renal function, indicating a need for longer follow-up of these patients.

The previously mentioned report from Mexico, in which the mortality rate was 83%, contained no data related to the need for dialysis (1). As previously mentioned, 36% of our 25 H1N1-infected patients with AKI required dialysis. This frequency of the need for dialysis is different from that reported in two studies of H1N1-infected patients in Argentina. In one of those studies, conducted at a general hospital, 89% of the patients required dialysis (14), compared with only 18% in the other study, which was conducted more recently and involved only critically ill patients (10). In the study conducted in Canada, the frequency of the need for dialysis was 11% and peaked on day 28 of hospitalization (9).

Mortality rates of 54.6% and 16%, respectively, have been reported among critically ill H1N1-infected patients in Argentina and Canada (9,10). In both of those studies, AKI was associated with higher mortality. In the study presented here, overall mortality was higher in the AKI group, and, more importantly, death occurred only among the AKI patients. This reflects the greater disease severity among AKI patients, as evidenced by the higher mean APACHE II scores in the AKI group. Mortality was associated with the frequency of the need for dialysis, which was 66% among nonsurvivors and 19% among survivors. The need for dialysis was also associated with mortality in the study conducted in Argentina, although not in the study conducted in Canada (9,10). It is known that fluid overload increases mortality among critically ill patients (15). In the study presented here, dialysis was initiated early (seven of the nine dialyzed patients received dialysis within 24 hours after the initial nephrology consultation) to avoid fluid overload. Nevertheless, hypervolemia was an indication for dialysis in eight of the nine patients requiring dialysis. Therefore, a nephrology consultation should probably have been requested sooner. It is interesting that a better RIFLE score at ICU admission was associated with higher mortality: seven of the nine nonsurviving AKI patients had not reached even the minimum “risk” score at ICU admission, showing that, despite appropriate treatment, those patients developed AKI after ICU admission. This is in agreement with the reported trend toward higher mortality when AKI develops during the ICU stay than when it is present at ICU admission (16). It is likely that it is early AKI, rather than postadmission AKI, that is associated with H1N1 infection.

In summary, a significant number of patients with severe H1N1 infection develop AKI, which increases mortality in such patients. In our study, death only occurred in AKI patients. It is likely that the development of AKI in H1N1-infected patients involves multiple mechanisms, including hemodynamic instability, hypoxia, and a direct effect of the H1N1 virus on the kidney. In H1N1-infected patients who develop AKI, renal function should be monitored after hospital discharge.

Disclosures

None.