Abstract
Background and objectives AKI is a clinical syndrome with various causes involving glomerular, interstitial, tubular, and vascular compartments of the kidney. Acute kidney disease (AKD) is a new concept that includes both AKI and the conditions associated with subacute decreases in GFR (AKD/non-AKI). This study aimed to investigate the correlation between AKI/AKD defined by clinical presentation and diffuse histologic criteria for acute abnormalities based on renal biopsy.
Design, setting, participants, & measurements All 303 patients who were histologically diagnosed as having acute tubular necrosis (ATN), acute tubulointerstitial nephritis, cellular crescentic GN, acute thrombotic microangiopathy, or complex lesions on renal biopsy from January 2009 to December 2011 were enrolled in the study. The 2012 Kidney Disease Improving Global Outcomes AKD/AKI definitions were applied to classify patients as follows: AKI, AKD/non-AKI, non-AKD, or unclassified.
Results A total of 273 patients (90.1%) met the AKD criteria; 198 patients (65.3%) were classified as having AKI according to serum creatinine (SCr) and urine output criteria. The urine output criteria added 4.3% to the SCr criteria and reclassified 6.7% of the AKI cases into higher stages. Of patients with ATN on pathology, 79.2% met AKI criteria; this was a higher percentage than for those who had other individual pathologic lesions (50%–64%). The major cause of not being defined as having AKI was a slower SCr increase than that required by the definition of AKI (98, 93.3%). Patients with AKI had more severe clinical conditions and worse short-term renal outcome than those in the non-AKI group.
Conclusions Diffuse, acute abnormality defined by renal biopsy and AKI defined by clinical presentation are two different entities. Most patients who have diffuse acute histologic findings met the criteria for AKD, whereas only two thirds met the definition of AKI.
Introduction
The diagnosis of AKI and its various stages, either by the RIFLE (risk, injury, failure, loss of function, ESRD) classification (1) or the Acute Kidney Injury Network (AKIN) criteria (2), has been validated in various hospitalized populations to have strong predictive value for both short-term and long-term outcomes (3). Recently, the Kidney Disease Improving Global Outcomes (KDIGO) clinical practice guideline for AKI proposed a modified definition of AKI based on RIFLE and AKIN criteria that makes up a new conceptual definition for acute kidney disease (AKD). AKD was defined to include clinical conditions of acute (AKI) or subacute decreases in GFR (i.e., GFR<60 ml/min per 1.73 m2 for <3 months or decrease in GFR by ≥35% or increase in serum creatinine [SCr] by >50% for <3 months) (4).
The AKI and AKD definitions describe clinical states that result from various causes and are characterized by various types of histologic damage. A kidney disease can be clinically defined as “acute” on the basis of the clinical course. Pathologically, it may manifest with “acute” lesions, such as cellular crescents, tubular necrosis, and leukocyte infiltrates. Nevertheless, the relationship between clinical acute course and pathologic acute lesions is not well understood.
To date, few studies have related clinical AKI or AKD diagnoses to parenchymal pathologic changes present on renal biopsy because in many cases the clinical designation of AKI or AKD is not informed by a kidney biopsy. In a previous pilot study, we studied 740 cases for which a renal biopsy was done in 2011. Altogether, 131 cases were defined as AKD (17.7%), of which 62 cases were defined as AKI (8.4%). Among these patients with AKD/AKI, the most common pathologic changes included acute tubulointerstitial nephritis (ATIN), acute tubular necrosis (ATN), cellular crescentic GN (CCGN), and thrombotic microangiopathy (TMA), accounting for 30%, 22%, 23%, and 9% of AKD cases and 29%, 29%, 26%, and 8% of AKI cases, respectively (L. Yang et al., unpublished data). These data suggest that these four broad types of renal parenchymal acute lesions account for 84% of AKD cases and 92% of AKI cases in our renal biopsy population. To better define the relationship between clinical AKI/AKD criteria and renal parenchymal acute lesions, we conducted a retrospective study of 2119 patients and evaluated clinical-pathologic correlations in 303 patients who had presented with evidence for acute lesions of ATIN, ATN, CCGN, or TMA documented on renal biopsy. We asked the following question: Of these patients with histologic evidence for acute injury, how many fulfilled the clinical criteria for AKI or AKD?
Materials and Methods
Patients
This study adhered to the principles of the Declaration of Helsinki and was approved by the Committee on Research Ethics of Peking University First Hospital. Patients who underwent renal biopsy from January 1, 2009, to December 31, 2011, in the Renal Division of Peking University First Hospital were screened. The renal biopsy specimens were examined by light microscopy with hematoxylin-eosin, periodic acid–Schiff, Masson’s trichrome, and periodic acid-methenamine silver staining; immunofluorescence; and electron microscopy. Patients who were identified as having ATN, ATIN, CCGN, and TMA were enrolled in the study (Figure 1). Patients from other medical departments, including intensive care units (ICUs), were not included according to the study design.
Flowchart of patient enrollment. ATN, acute tubular necrosis; ATIN, acute tubulointerstitial nephritis; CCGN, cellular crescentic GN; G, glomeruli; TIN, tubulointerstitial nephritis; TMA, thrombotic microangiopathy.
Pathologic Diagnostic and Inclusion Criteria
The indication for kidney biopsy included rapidly progressive GN; acute nephritic syndrome; unexplained acute renal failure; nephrotic syndrome (especially in patients older than age 40 years); and non-nephrotic proteinuria of more than 1 g/d or with glomerular hematuria, decreased renal function, or clinical or serologic evidence of a systemic disease.
CCGN.
At least 50% of the glomeruli had cellular crescents, and ≤40% of the glomeruli were sclerotic, which included global glomerulosclerosis and fibrous crescent (5,6). A cellular crescent was defined as an extracapillary cell proliferation with cells occupying ≥10% of the lesion, including both segmental and circumferential involvement (7).
ATN.
Patients were included when any of the following was found: loss of tubular brush border, tubular necrosis, intraluminal proteinaceous cellular debris or casts, tubular dilatation with flattening of tubular epithelium, or proliferation of the tubular cells (8) involving ≥50% of the area of biopsy specimens.
ATIN.
At least 50% of the parenchymal area had interstitial inflammation, with <25% of the area having interstitial fibrosis (8).
TMA.
Thrombosis of the glomerular capillaries and arterioles and onion skin–like thickening of small arteries were detected under light microscopy with ≤40% of the glomeruli sclerotic. Widening of the subendothelial area was detected by electron microscopy (8).
Complex Lesions.
A combination of at least two types of the above-mentioned categories.
Scoring.
Scores were developed for classifying the tubular brush border loss, necrosis, and atrophy; interstitial edema; inflammation; and fibrosis according to the Banff working classification (9,10). A 0–4+ scale was applied as follows: 0=no lesion, 1+=<25%, 2+=>25%–50%, 3+=>50%–75%, and 4+=>75% of parenchyma affected by the lesion.
Clinical Data Collection and Grouping
Disease history, age, sex, weight, BP, underlying diseases, dialysis and urine volume, hemoglobin, SCr, and the length of hospital stay and costs were recorded. Patients were classified into clinical groups of AKI, AKD/non-AKI, non-AKD, or unclassified according to the 2012 KIDGO AKI SCr and urine output (UO) criteria and AKD definition (3). AKD is a broader concept that comprises both AKI and AKD/non-AKI.
AKI Group.
The diagnostic criteria for AKI were an abrupt increase in SCr by ≥0.3 mg/dl within 48 hours or to ≥1.5 times baseline that occurred within the prior 7 days and/or a decline of UO to ≤12 ml/kg per 24 hours for 24 hours. A modified 24-hour UO criterion was applied because hourly UO was not available in these non-ICU patients. Patients with UO≤12 ml/kg per 24 hours for 24 hours were allocated to AKI stage 2; those who had UO≤7.2 ml/kg per hours for 24 hours were defined as AKI stage 3. Baseline SCr was defined as the lowest value among the following: the SCr value that was measured at a time closest to admission within the prior 1 year, the minimum SCr value within 3 months before admission, or the minimum SCr value during the hospital stay.
In all, 262 patients had baseline SCr, of which 93 patients had no baseline CKD, 134 patients were at CKD stage 1, 16 patients were at stage 2, and 19 patients were at stage 3. In 34 patients who had no records of SCr tests with no suspected CKD, a back-estimation of baseline SCr was made from the Modification of Diet in Renal Disease formula, assuming an eGFR of 75 ml/min per 1.73 m2 (4).
AKD/Non-AKI Group.
Patients who had developed a GFR<60 ml/min per 1.73 m2 or had a decrease in GFR by ≥35%, or had an increase in SCr by >50% within 3 months before admission (4), but did not meet the AKI criteria were placed in the AKD/non-AKI group.
Non-AKD Group.
Patients who had GFR≥60 ml/min per 1.73 m2 or developed GFR<60 ml/min per 1.73 m2 over more than 3 months, presenting a slower SCr rise than that required by AKD definition, were classified into the non-AKD group.
Unclassified Group.
Patients who had GFR<60 ml/min per 1.73 m2, with history of proteinuria and/or hematuria of >3 months, but had no records of SCr measurements could not be identified as having acute or chronic kidney insufficiency.
Renal Outcome for Patients with AKD at Hospital Discharge
An eGFR was calculated using the CKD-Epidemiology Collaboration equation (11,12), from which CKD stage was assessed at discharge to evaluate renal outcome. CKD stages 1–2 at discharge was defined as a good outcome; CKD stages 3–5 was regarded as a poor outcome.
Statistical Analyses
SPSS software, version 16.0 (IBM, Chicago, IL), was used for statistical analyses. Quantitative data are expressed as mean±SD or median with 25th, 75th percentiles. Categorical data are expressed as frequency (percentage). For comparison of clinical and pathologic parameters among different groups, nonparametric variables were analyzed by the Mann–Whitney test or the Kruskal–Wallis test, and categorical variables were analyzed by Pearson two-tailed test or Fisher exact test, whenever appropriate. To determine the association of clinical factors, such as age, sex, comorbid conditions, oliguria, systolic BP, baseline SCr, RRT, AKI stage, hemoglobin, nephrotic syndrome, and pathologic changes, with AKI classification and renal outcome of patients with AKD at time of discharge, we used univariate logistic regression analysis followed by multivariate logistic regression with stepwise backward inclusion of variables to minimize the number of covariates in the model. P values were two sided, and P<0.05 was considered to indicate statistical significance.
Results
Patient Enrollment, Clinical Diagnosis, and Grouping
Of 2119 biopsy specimens, 303 cases (159 men and 144 women, age 45.7±17.1 years) met the pathologic enrollment criteria (Figure 2). The renal pathologic diagnoses included ATIN in 107 cases (35.3%), CCGN in 89 (29.4%), ATN in 77 (25.4%), TMA in 22 (7.3%), and complex lesions in 8 (2.6%). Altogether 273 patients (90.1%) met AKD criteria, of whom 198 patients (65.3% of total) were classified as having AKI, with 32 at stage 1 (16.2%), 14 at stage 2 (7.1%), and 152 at stage 3 (76.8%). Seven patients (2.3%) were grouped as unclassified because their disease course was difficult to determine.
Clinical grouping of patients with renal parenchymal acute lesions.
Effect of UO Criteria on AKI Diagnosis and Staging
On the basis of SCr criteria, 185 patients (61.1%) were identified as having AKI. Seventy-one (23.4%) met the UO criteria for AKI, of whom 13 (4.3%) did not meet SCr criteria. These 13 patients, including 7 who had inadequate SCr records and 6 who showed relatively slow increases in SCr levels, presented with rapid decreases in UO and therefore were diagnosed and staged by UO criteria. In addition, implementing UO criteria increased AKI stage over that defined by the SCr criteria in another 13 patients (Table 1). All the 19 patients who did not meet SCr criteria because of a relatively slow increase in SCr or were graded lower by the SCr criteria presented with nephrotic syndrome. Body weight increased by 10.5±7.6 kg (median increase, 8.25 [interquartile range, 5.0–15.0 kg]) during the disease course, with significantly decreased serum albumin of 1.52±0.32 g/dl.
AKI diagnosis and staging by serum creatinine/urinary output criteria
Applicability of AKD/AKI Definition in Various Pathologic Groups
AKI definition identified all patients with complex lesions and more cases from the ATN group (79.2%) than any other individual pathologic groups (50.0%–64.0%; P=0.02) (Figure 3). The AKD criteria characterized a large percentage of patients with ATN (94.8%), ATIN (93.5%), and CCGN (88.7%). AKI stage did not significantly differ among different pathologic groups, except for a higher proportion of RRT prescription in the CCGN group (P<0.001).
Acute diffuse abnormality defined by renal biopsy and AKI defined by clinical presentation are two different entities. (A) No significant difference was seen in the composition of various pathological changes among different clinical groups. (B) AKI definition identified all patients with complex lesions and more cases from the ATN group than any other individual pathological groups. In addition, 2.2% of patients with cellular crescent GN (CCGN) and 22.7% of patients with thrombotic microangiopathy (TMA) were grouped as unclassified. AKD, acute kidney disease.
Reasons for Not Meeting the AKI Definition
We then investigated the reasons for not meeting the AKI definition in a sizable percentage (34.7% [105 of 303]) of patients. A slower SCr increase than that required by the AKI definition was the major cause (93.3% [98 of 105]). Another reason was unavailable disease records (7 patients [6.7%], unclassified group).
We further looked into factors that might contribute to the slower increase in SCr. Patients who had the slowest SCr increase (non-AKD group) were younger and had milder renal dysfunction than patients who were identified as having AKD/non-AKI or AKI (Table 2). Unfortunately, by multiple logistic regression analysis, we could not define any factors significantly associated with AKI classification. None of the pathologic lesion scores were related to AKI diagnosis. This is understandable because we included only patients with severe tissue damage (i.e., >50% of the parenchymal area was affected).
Clinical characteristics of patients in different clinical groups
Patients Classified with AKI Had the Highest Severity of Injury
Patients in the AKI group presented the highest SCr value at biopsy and at peak, the highest hospital costs, the longest hospital stay, and the worst renal outcome at discharge compared with those in the groups of AKD/non-AKI or non-AKD (Table 2). Patients with AKI stage 1 and stage 2 had similar clinical features (Table 3). In total, 113 patients required RRT (Table 2), presenting with the highest systolic BP, the highest SCr value, the highest hospital costs, the longest hospital stay, and the worst renal outcome; 45.3% remained on dialysis at discharge (Tables 3 and 4). Two patients who required RRT in the AKI group died: 1 of pulmonary infection and 1 of cerebral hemorrhage. This relatively low mortality rate in RRT recipients (1.8% [2 of 113]) is, at least in part, due to the fact that patients in the ICU were excluded.
Clinical characteristics of patients in various AKI stages
Renal outcome of patients with AKI who required RRT in various pathologic groups
We then investigated the factors related to renal outcome at time of discharge. On multiple logistic regression analysis, age (odds ratio [OR], 1.69 per 10-year increment; 95% confidence interval [95% CI], 1.28 to 2.22; P<0.001), underlying cardiovascular diseases, diabetes, or hypertension (OR, 2.37; 95% CI, 1.09 to 5.18; P=0.03), baseline SCr (OR, 1.08; 95% CI, 1.04 to 1.12; P<0.001), and RRT (OR, 8.41; 95% CI, 2.78 to 25.43; P<0.01) were independent risk factors for worse renal outcome at discharge.
Discussion
The current AKI definitions and staging criteria mainly derive from patients in the ICU, in whom the predominant cause of AKI was related to ischemia, toxins, and severe sepsis (13–15). The present study aimed to assess KDIGO AKI/AKD definitions in patients with indications for kidney biopsy who had various documented diffuse acute pathologic kidney lesions.
Of the 303 patients with diffuse parenchymal acute renal pathologic injury, only 198 (65.3%) were classified as having clinical AKI. The reliability of clinical AKI recognition varied with different kinds of pathologic lesions, with a much better concordance in ATN than in ATIN, CCGN, or TMA. Further analyses revealed that the major reason for not meeting the AKI criteria was the slower decrease in GFR, reflecting underlying pathophysiologic processes in various pathologic injuries. It is not surprising that patients who most commonly met the AKI criteria most often had abrupt loss of GFR after acute tubular injury, where the definition was derived. In cases of ATIN, CCGN, and TMA, however, where maladaptive immunity plays major roles in the tissue damage, the loss of GFR was less aggressive but more progressive than in ATN cases and thus was not in accordance with AKI criteria, which focuses more on the speed than the length of the disease course. In the present study, 20.8% of the patients with ATN were not classified as having AKI because of a slower GFR loss as well. Our study did not include patients in the ICU, in whom the cause and pathogenesis of ATN are generally complex and occur in the setting of severe clinical conditions (16). ATN in the non-ICU unit, such as the cases in our study, is usually induced by causes such as drugs or milder ischemia and, thus, may be less severe and less aggressive compared with ATN cases in the ICU.
Another possible reason that some patients with diffuse acute pathology did not meet the AKI criteria may be the presence of “renal reserve” (17), the “resting” extra excretory capacity that the kidney can call upon in times of stress. When the kidney is damaged and filtration capacity is lost, this unused reserve would be depleted before the resting GFR falls (17). Vaidya et al. demonstrated that SCr value does not increase until the renal parenchyma is significantly injured (18). This can easily explain why younger patients tended not to meet AKI/AKD criteria in our study. In addition, patients who had nephrotic syndrome and significant edema were less likely to meet AKI criteria. In these patients, the increase in SCr could be slowed by dilution. These results re-emphasize the limitation of SCr in detecting AKIs.
UO criteria were important for fine-tuning the diagnosis and staging of AKI in our cohort. Recent studies have shown that UO is a sensitive and early marker for AKI in patients in the ICU and has additional roles in AKI staging (19,20). There is still controversy, however, about the application of UO criteria, especially in non-ICU patients. In the present study, a modified 24-hour UO criterion was applied; this added 4.3% to the SCr criteria and reclassified 6.7% of the patients with AKI into higher stages. It was of particular use for the patients who had insufficient SCr records and those who had nephrotic syndrome.
Because the major reason for not meeting AKI criteria was the less aggressive GFR loss, it is not surprising that a much higher proportion of patients (92.1%) were classified as having AKD, with good agreement among the groups with various pathologic lesions. Furthermore, even though the patients in the AKD/non-AKI group presented with slower GFR declines, they had similar durations of hospital stay, total costs, and renal outcome at discharge compared with those who had AKI stages 1 and 2. This study, to our knowledge, is the first application and validation of the KDIGO AKI/AKD definitions in patients with acute parenchymal kidney diseases. Our results suggest that the operational AKD definition could identify kidney injury at an earlier stage, especially in patients with pathologic lesions other than ATN.
Our study has limitations related to the retrospective observational design in a single center. The major limitation is the selection bias that stems from retrospectively evaluating clinical-pathologic correlations in patients who had presented with evidence for acute lesions of ATIN, ATN, CCGN, or TMA on renal biopsy. This design excluded information from patients who were too sick to undergo biopsy or had contraindications for renal biopsy. Moreover, these patients are more likely to have met the AKI definition. Although we did consider the limitation of using SCr to detect acute pathologic injury, it is important to also highlight the limitation of using the kidney biopsy specimens for the same (e.g., because of inadequate tissue sampling). An additional limitation of our study is that the pathologic criteria, 40% and 25% cutoffs for sclerotic and interstitial fibrosis, may not completely exclude chronic processes. The use of the Modification of Diet in Renal Disease formula to back-estimate baseline SCr may have also limited the accuracy because it has not been validated in the Chinese population. The short-term outcomes evaluated by our study is another limitation. Many of the preceding limitations may be addressed by a multicenter, prospective investigative study conducted over a longer term.
In conclusion, the KDIGO definition of AKD detects most patients with a broad range of acute renal parenchymal histologic injuries on kidney biopsy. The AKI and AKD/non-AKI clinical courses reflect different type of renal abnormality, and the AKI criteria help distinguish patients with more severe renal dysfunction and worse outcomes.
Disclosures
None.
Acknowledgments
This work was supported by National Natural Science Foundation of China grant 81070549 and Beijing Training Program for the Talents 20110009001000002 to L.Y.
Footnotes
Published online ahead of print. Publication date available at www.cjasn.org.
- Received June 11, 2013.
- Accepted March 13, 2014.
- Copyright © 2014 by the American Society of Nephrology