Article

Epidemiology and Natural History of the Cardiorenal Syndromes in a Cohort with Echocardiography

| Thomas A. Mavrakanas,*† Aisha Khattak,*‡ Karandeep Singh,§ and David M. Charytan*

Abstract Background and objectives It is unknown whether echocardiographic parameters are independently associated with the cardiorenal syndrome. No direct comparison of the natural history of various cardiorenal syndrome types *Renal Division, has been conducted. Brigham and Women’s Hospital, Harvard Design, setting, participants, & measurements Our retrospective cohort study enrolled adult patients with at least Medical School, one transthoracic echocardiography between 2004 and 2014 at a single health care system. Information on Boston, Massachusetts; comorbidities was extracted using condition-specific diagnostic codes. All-cause mortality was the primary † – General Internal outcome among patients with cardiorenal syndrome types 1 4. and stroke were the Medicine Division, secondary outcomes. Geneva University Hospitals, Geneva, Results In total, 30,681 patients were included, and 2512 (8%) developed at least one of the cardiorenal syndromes: Switzerland; and ‡Division of Renal 1707 patients developedan acute formofthesyndrome (type1or3),128patients developed type2, and677patients Disease and developed type 4. In addition, 16% of patients with type 2 and 20% of patients with type 4 also developed an , acute cardiorenal syndrome, whereas 14% of patients with acute cardiorenal progressed to CKD or chronic heart Department of Internal failure. Decreasing left ventricular ejection fraction, increasing pulmonary artery pressure, and higher right Medicine, The ventricular diameter were independently associated with higher incidence of a cardiorenal syndrome. Acute University of Texas Health Science Center, cardiorenal syndrome was associated with the highest risk of death compared with patients with CKD without Houston, Texas; cardiorenal syndrome (hazard ratio,3.13;95% confidence interval, 2.72 to 3.61; P,0.001). Patients with cardiorenal §Departments of fi Learning Health type 4 had better survival than patients with acute cardiorenal syndrome (hazard ratio, 0.48; 95% con dence | interval, 0.37 to 0.61; P,0.001). Patients with acute cardiorenal syndrome and type 4 had increased risk of Sciences and Internal Medicine, University myocardial infarction and stroke compared with patients with CKD without cardiorenal syndrome. of Michigan Medical School, Ann Arbor, ConclusionsUp to 19% of patients with a chronic form of cardiorenal syndrome will subsequently develop an acute Michigan syndrome. Development of acute or type 4 cardiorenal syndrome is independently associated with mortality, the acute form having the worst prognosis. Correspondence: Clin J Am Soc Nephrol 12: 1624–1633, 2017. doi: https://doi.org/10.2215/CJN.04020417 Dr. Thomas A. Mavrakanas, Renal Division, Brigham and Women’s Hospital, Introduction Prognosis seems to be different if the deterioration 1620 Tremont Street, The cardiorenal syndrome encompasses the various Boston, MA 02120. in kidney function is spontaneously observed on admis- Email: tmavrakanas@ interactions between the heart and the kidneys in the sion or is a -related effect, the latter conferring partners.org or acute or chronic setting. Several studies have attempt- better prognosis when associated with hemoconcen- tmavrakanas@gmail. ed to present different components of the syndrome, tration (7). com the epidemiology, and the natural history of the heart- In addition, an AKI episode may lead to subsequent kidney interaction (1). The most commonly used deterioration in heart function or an acute cardiac classification of cardiorenal syndrome is the one event. The term cardiorenal syndrome type 3 has been proposed by Ronco and colleagues (2–4), which breaks suggested for this form (2,8). Contrast-induced ne- it down into five categories on the basis of the direction phropathy or the acute glomerulonephritides fall into and chronicity of interaction between the heart and this category, with volume overload presumed to be kidneys. the main pathophysiologic mechanism underlying the Most studies have focused on acute cardiac events, secondary cardiac events. mainly acutely decompensated (HF), Outside the acute setting, several studies, thor- leading to AKI. This form of the syndrome is also oughly reviewed by Bagshaw et al. (1), have identified known as cardiorenal type 1 (2,5). Worsening kidney that chronic heart disease and CKD commonly co- function in patients admitted for acute HF has been exist. In many patients, it is not possible to determine associated with an increased risk of death (6). How- which entity is primary versus secondary. In a pooled ever, it is unclear whether worsening kidney func- analysis of two longitudinal cohort studies, patients tion is a mediator or a marker of worse outcomes. with baseline were at higher

1624 Copyright © 2017 by the American Society of www.cjasn.org Vol 12 October, 2017 Clin J Am Soc Nephrol 12: 1624–1633, October, 2017 Epidemiology and Natural History of Cardiorenal Syndromes, Mavrakanas et al. 1625

risk of developing compared with patients The following baseline laboratory data were extracted without cardiovascular disease at baseline (9), suggesting (the values closest to the echocardiogram date within an that cardiovascular disease can be independently associ- interval of 1 year before to 3 months after the echocardio- ated with progressive decline in kidney function. This entity gram): hemoglobin, albumin, glycated hemoglobin, calcium, has been described as cardiorenal syndrome type 2 (2,10). phosphorus, parathyroid hormone, urine albumin-to-creatinine The inverse association has also been studied. In several ratios, and brain natriuretic peptide levels. large retrospective cohorts, CKD was associated with We determined baseline medication use by matching increased cardiovascular event rates and all-cause mortal- the most recent medication list with an exhaustive list of ity (11,12). The association was strongest at more advanced generic and brand names in these categories: angiotensin- CKD stages (12). This form of interaction is known as converting enzyme inhibitors, angiotensin receptor block- cardiorenal syndrome type 4 (2,13). Finally, cardiorenal ers, b-blockers, , aspirin, clopidogrel, and warfarin. syndrome type 5 is attributed to situations where a systemic If the medication was prescribed before the date of the condition affects both the heart and kidneys (14). echocardiogram, the patient was considered to be on that Despite this evidence, understanding the complex heart- medication at baseline. kidney interactions remains incomplete. Reported inci- All echocardiogram reports were acquired and record- dence rates significantly vary due to the heterogeneity of edusinganAmericanSocietyofEchocardiography– the studied populations. Few studies have clearly discrim- recommended scanning protocol with standard techniques. inated between the primary disease and the secondary The following parameters were extracted using pattern condition, especially in the chronic forms of the syndrome. matching: left ventricular ejection fraction (LVEF), left ven- In particular, whether echocardiographic data are inde- tricular (LV) diastolic diameter, LV mass index (corrected for pendently associated with the occurrence of the syndrome body surface area), left atrium diameter, pulmonary arterial has not been well studied, and no direct comparison of the pressure, right atrial pressure, regional wall motion abnor- natural history of these forms has been conducted. malities, right ventricular (RV) systolic function (preserved or We studied a large echocardiographic cohort to describe reduced), and hypertrophy or dilation (presence or absence). the prevalence of different types of cardiorenal syndrome Patients with LVEF.50% were considered to have preserved and compare mortality and myocardial infarction (MI) and ejection fraction. stroke rates in patients with and without cardiorenal Echocardiographic parameters and medication use were syndrome. derived via processing of the free text reports and text- based fields with prescription names. At least 100 patients were manually reviewed for each medication and each Materials and Methods echocardiographic parameter to assess extraction accuracy. Study Population This was a retrospective cohort study enrolling adult Cardiorenal Syndrome Definitions patients referred for transthoracic echocardiography with Acute cardiorenal syndromes (type 1 or 3) were consid- tissue Doppler imaging at a Partners health care facility ered as a single group defined by the presence of codes for between January 1, 2004 and January 6, 2014. The Partners both HF decompensation and AKI (584.5–584.9) during the Institutional Review Board approved the study. Patients same admission. Type 2 cardiorenal syndrome was diagnosed $18 years old without history of kidney or heart transplant in patients with HF at baseline and an eGFR$60 ml/min per and with at least one echocardiogram and one creatinine 1.73 m2 who subsequently doubled their creatinine values, value during the 365 days before and the 90 days after the were started on maintenance dialysis, or had a kidney echocardiogram were included. The average creatinine transplant. Type 4 cardiorenal syndrome was diagnosed in value was calculated for each patient using all of the patients without baseline HF or HF within 30 days from the available creatinine values during the 365 days before and first echocardiogram, with an eGFR,60 ml/min per the 90 days after the index echocardiogram. Baseline eGFR 1.73 m2, who were subsequently admitted for decompen- was then calculated with the CKD Epidemiology Collab- sated HF. Patients with more than one type of cardiorenal oration equation (15). syndrome were classified by their first presentation.

Data Collection Control Groups The Research Patient Data Registry (RPDR) was used to We constructed two control groups. The first (HF with retrieve encounters, medications, laboratory tests, and stable kidney function) included patients with baseline HF echocardiogram results. The RPDR is a centralized data and an eGFR$60 ml/min per 1.73 m2 who did not develop registry gathering clinical information from the Partners an acute cardiorenal syndrome between the first echocar- health care facilities, which include five teaching hospitals. diogram and the day of the last clinical visit or laboratory Information on comorbidities was extracted using the test, did not double their creatinine, and did not start ninth revision of the International Classification of Diseases dialysis or have a kidney transplant. The second (CKD with (ICD-9) diagnostic codes. To be considered as having the stable heart function) included patients with baseline respective comorbidities, each patient was required to have a eGFR,60 ml/min per 1.73 m2 without HF who did not relevant inpatient or outpatient diagnosis before the echocar- develop an acute cardiorenal syndrome between the first diogram. Because many echocardiograms were performed echocardiogram and the last clinical visit or laboratory test during a first admission for decompensated HF, any patient and were not admitted for decompensated HF in the same with an HF diagnostic code in the first month after the time period. At least 12 months of follow-up were required echocardiogram was considered to have HF at baseline. in both groups. 1626 Clinical Journal of the American Society of Nephrology

Figure 1. | Flowchart of the study population according to presence or absence of CKD and congestive heart failure (CHF) at baseline and type of cardiorenal syndrome during follow-up. Tx, transplant.

Outcomes regression. For this analysis, group assignment was de- All-cause mortality was the primary outcome; nonfatal termined on the basis of events during the first year of MI and stroke were the secondary outcomes. Clinical follow-up, and survival was credited from the time of the outcomes were identified using ICD-9 diagnostic codes. landmark, arbitrarily placed at 1 year from the first MI was considered fatal when death occurred within 30 echocardiogram. Allocation to the cardiorenal or control days. The RPDR dataset is continuously matched to the Social Security Death Index, a national database including all deaths in the United States. Patients were censored at the day of the last clinical visit or laboratory test.

Statistical Analyses Baseline characteristics were compared using the chi- squared test and one-way ANOVA or the Mann–Whitney test as appropriate. The Bonferroni correction for post hoc tests was applied. To identify which covariates contribute to the develop- ment of any form of the syndrome, we used Cox pro- portional hazards regression. The analysis included all patients. A second analysis included only those with HF and/or CKD at baseline to assess whether associations differed in those with impaired kidney or heart function at baseline. The following covariates were included in the baseline model: age, sex, race, hypertension, , HF, (CAD), and baseline eGFR. We then examined whether the echocardiographic parameters Figure 2. | Substantial overlap is observed between the different were associated with the cardiorenal syndrome. The cohort forms of the cardiorenal syndrome. Twenty patients presented was divided into quartiles for pulmonary artery pressure. with cardiorenal syndrome type 2 and subsequently developed an For LVEF, the following cutoffs were used: ,25%, 25%– acute cardiorenal syndrome (16% of patients with a first presentation 39%, 40%–54%, and $55%. RV dilation was used as a of cardiorenal syndrome type 2), whereas 78 patients presented with binary parameter. These parameters were selected to reflect an acute cardiorenal syndrome and subsequently developed car- LV systolic function and RV pressure and volume over- diorenal syndrome type 2 (5% of patients with a first presentation of acute cardiorenal syndrome). Similarly, 135 patients presented load. The proportional hazards assumption was verified with cardiorenal syndrome type 4 and subsequently developed an using graphical methods. acute cardiorenal syndrome (20% of patients with a first presentation We compared mortality, nonfatal MI, and stroke rates of cardiorenal syndrome type 4), whereas 164 patients presented with across the different cardiorenal groups (acute, type 2, and an acute cardiorenal syndrome and subsequently developed car- type 4) and the two control groups using Kaplan–Meier diorenal syndrome type 4 (10% of patients with a first presentation of plots, the log rank test, and Cox proportional hazards acute cardiorenal syndrome). Clin J Am Soc Nephrol 12: 1624–1633, October, 2017 Epidemiology and Natural History of Cardiorenal Syndromes, Mavrakanas et al. 1627

group was also determined at this time point for this cardiorenal syndrome type 4. Also, 3072 patients had stable analysis. Patients who had an event before the landmark HF and were classified in the first control group, whereas were excluded. Hazard ratios (HRs) were adjusted for age, 2239 patients had stable CKD and were classified in the sex, race, hypertension, diabetes, and CAD. second control group (Figure 1). SPSS (version 20.0; SPSS Inc., Chicago, IL) and Stata There was substantial overlap during follow-up, with the (version 14 IC; College Station, TX) were used. P,0.05 was same patient often developing both an acute and a chronic considered statistically significant. form of the syndrome: 20 of 128 patients (16%) with cardiorenal type 2 and 135 of 677 patients (20%) with cardiorenal type 4 subsequently developed an acute form. Acute cardiorenal Results syndrome preceded type 2 in 5% of acute patients (78 of The cohort included 31,173 patients. A total of 492 1707) and type 4 in 10% of the acute patients (164 of 1707) patients had history of kidney transplant and were excluded (Figure 2). Patients with an acute form had, on average, 2.9 from this analysis. Echocardiograms were performed in acute events during follow-up (median, 2; interquartile 16,968 inpatients (55%) and 13,713 outpatients (45%). Of range, 1–3). the 30,681 patients, 2512 (8.2%) developed at least one of the Patients with cardiorenal syndrome had different base- cardiorenal syndrome types; 1707 patients developed an line characteristics from stable patients with CKD (Table 1). acute form of the cardiorenal syndrome (type 1 or 3), 128 Compared with stable patients with HF, they were older; developed cardiorenal syndrome type 2, and 677 developed had higher prevalence of diabetes; had worse baseline

Table 1. Baseline characteristics of patients with any form of the cardiorenal syndrome and the control groups

Characteristic Any Cardiorenal Control 1 (Stable HF) Control 2 (Stable CKD)

N 2512 3072 2239 Age, yr 68614a,b 62615 71613 Men 1332 (53%)a,b 1746 (57%) 1037 (46%) Black 344 (14%)a,b 326 (11%) 217 (10%) Hypertension 1474 (59%)b 1753 (57%) 1468 (66%) Diabetes mellitus 840 (33%)a,b 761 (25%) 516 (23%) CAD 1504 (60%)b 1763 (57%) 1064 (48%) ACEI or ARB 1187/2310 (51%)a,b 1299/2889 (45%) 857/2050 (42%) b-Blocker 1494/2310 (65%)a,b 1963/2889 (68%) 1109/2050 (54%) 1210/2310 (52%)b 1431/2889 (50%) 944/2050 (46%) Loop 1240/2310 (54%)b 1546/2889 (54%) 387/2050 (19%) Spironolactone 203/2310 (9%)a,b 157/2889 (5%) 36/2050 (2%) Aspirin 1332/2310 (58%)b 1691/2889 (59%) 929/2050 (45%) Warfarin 506/2310 (22%)b 612/2889 (21%) 251/2050 (12%) Creatinine, mg/dl 1.26 (1.00–1.62)a,b 0.87 (0.75–1.01) 1.30 (1.11–1.56) eGFR, ml/min per 1.73 m2 54624a,b 85616 46612 UACR, mg/g 105 (25–545) (n=225)a,b 22 (10–83) (n=152) 31 (12–180) (n=215) Hb, g/dl 11.862.1 (n=1365)a,b 12.462.2 (n=2098) 12.162.0 (n=1232) Albumin, g/dl 3.860.6 (n=2034)b 3.860.6 (n=2498) 3.960.6 (n=1648) HbA1c, % 6.7 (5.8–7.8) (n=124) 6.1 (5.7–7.4) (n=120) 6.2 (5.7–7.1) (n=85) BNP, pg/ml 385 (157–791) (n=805)a,b 244 (83–547) (n=864) 134 (58–290) (n=199) Preserved EF 1254/2310 (54%)b 1515/2855 (53%) 1781/2113 (84%) LVEF, % 55 (35–60) (n=2310)b 55 (35–60) (n=2855) 60 (55–60) (n=2113) LVd, cm 5.061.0 (n=2108)b 5.061.0 (n=2597) 4.460.7 (n=1759) LVMi, g/m2 110 (90–136) (n=601)a,b 102 (80–129) (n=927) 91 (75–110) (n=473) LAd, cm 4.360.9 (n=2054)a,b 4.160.8 (n=2509) 3.960.7 (n=1720) PAP, mm Hg 29 (23–39) (n=1621)a,b 26 (20–34) (n=1948) 25 (20–31) (n=1380) . b RAP 6cmH2O 133/692 (19%) 184/1190 (16%) 39/702 (6%) Preserved RV function 1650/2003 (82%)b 2149/2539 (85%) 1702/1757 (97%) Dilated RV 408/2092 (20%)a,b 410/2665 (15%) 135/1799 (8%) RV hypertrophy 75/976 (8%)a,b 63/1361 (5%) 25/1005 (3%) RWMA 207/1083 (19%)b 303/1441 (21%) 100/1334 (8%)

The Bonferroni correction for post hoc tests was applied. HF, heart failure; N, number of patients in each group; CAD, coronary artery disease;ACEI,angiotensin-convertingenzyme inhibitor; ARB,angiotensinreceptor blocker; UACR, urine albumin-to-creatinine ratio;n, number of patients with available data; Hb, hemoglobin; HbA1c, glycated hemoglobin; BNP, brain natriuretic peptide; EF, ejection fraction; LVEF, left ventricular ejection fraction; LVd, left ventricular diastolic diameter; LVMi, left ventricular mass index (corrected for body surface area); LAd, left atrium diameter; PAP, pulmonary arterial pressure; RAP, right atrial pressure; RV, right ventricular; RWMA, regional wall motion abnormalities. aP,0.05 versus control group 1. bP,0.05 versus control group 2. 1628 Clinical Journal of the American Society of Nephrology

kidney function; had higher LV mass index, left atrial Table 3 shows the results of the Cox proportional hazards diameter, or pulmonary artery pressure; and had higher model for the development of the cardiorenal syndrome. prevalence of dilated or hypertrophic RV. Echocardiographic parameters were independently associ- Supplemental Table 1 and Table 2 compare the patients ated with the development of the cardiorenal syndrome. with different types of the cardiorenal syndrome. Patients For the mortality analysis, 421 patients were identified with the acute form of the syndrome were different from with acute cardiorenal syndrome (type 1 or 3), ten patients patients with cardiorenal syndrome type 4 in most baseline were identified with type 2, and 183 patients were iden- characteristics. Compared with patients with cardiorenal tified with type 4. A total of 3436 patients were allocated to syndrome type 2, they were older; had higher prevalence of the first control group (stable HF), and 2768 patients were diabetes or CAD; had worse baseline kidney function; had allocated to the second control group (stable CKD). higher LV diameter, LV mass index, and left atrial di- Supplemental Table 2 shows the number of events in ameter; and had more regional wall motion abnormalities. each group. The survival distributions were significantly The two chronic forms also looked different: patients with different (Figure 3). Table 4 shows the HRs of the different type 2 were younger, had better kidney function but worse cardiorenal groups adjusted for age, sex, race, and history of LV or RV systolic function, had higher LV and RV hypertension, diabetes, and CAD. The acute form was diameters, and had higher prevalence of elevated right associated with the highest HR for mortality. Patients with atrial pressure. These differences can be explained, at least type 4 had better prognosis than patients with the acute in part, by the definition of these two types. form of the syndrome.

Table 2. Comparison of baseline characteristics among the different cardiorenal groups

Characteristic Acute Cardiorenal (Type 1 or 3) Cardiorenal Type 2 Cardiorenal Type 4

N 1707 128 677 Age, yr 66614a,b 59615b 73612 Men 968 (57%)b 63 (49%) 301 (45%) Black 263 (15%)b 16 (13%) 65 (10%) Hypertension 1011 (59%) 70 (55%) 393 (58%) Diabetes 612 (36%)a,b 30 (23%) 198 (29%) CAD 1073 (63%)a,b 62 (48%) 369 (55%) ACEI or ARB 876/1588 (55%)a,b 43/121 (36%) 268/601 (45%) b-Blocker 1067/1588 (67%)b 76/121 (63%) 351/601 (58%) Statin 877/1588 (55%)a,b 47/121 (39%) 286/601 (48%) Loop diuretics 1016/1588 (64%)b 72/121 (60%)b 152/601 (25%) Spironolactone 183/1588 (12%)b 9/121 (7%)b 11/601 (2%) Aspirin 965/1588 (61%)b 61/121 (50%) 306/601 (51%) Warfarin 386/1588 (24%)b 25/121 (21%) 95/601 (16%) Creat, mg/dl 1.25 (1.00–1.64)a,b 0.83 (0.70–1.01)b 1.33 (1.12–1.68) eGFR, ml/min per /1.73 m2 56624a,b 86617b 43613 UACR, mg/g 109 (30–540) (n=157) 174 (9–1211) (n=12) 68 (14–499) (n=56) Hb, g/dl 11.862.2 (n=975) 11.562.1 (n=70) 11.862.1 (n=320) Albumin, g/dl 3.760.6 (n=1392)a,b 3.460.7 (n=113)b 3.960.6 (n=529) HbA1c, % 6.7 (5.7–7.9) (n=83) 7.3 (6.0–8.3) (n=7) 6.7 (6.1–7.8) (n=34) BNP, pg/ml 419 (176–873) (n=642)b 273 (123–709) (n=50) 216 (90–510) (n=113) Preserved EF 734/1575 (47%)b 63/114 (55%)b 457/621 (74%) LVEF, % 50 (30–60) (n=1575)b 55 (40–60) (n=114)b 55 (50–60) (n=621) LVd, cm 5.161.1 (n=1440)a,b 4.960.9 (n=111)b 4.660.8 (n=557) LVMi, g/m2 113 (93–143) (n=440)a,b 97 (76–108) (n=32) 105 (81–126) (n=129) LAd, cm 4.460.9 (n=1534)a,b 4.160.8 (n=89) 4.160.8 (n=431) PAP, mm Hg 30 (23–40) (n=1143)b 30 (21–40) (n=72) 28 (23–36) (n=406) . b b RAP 6cmH2O 112/490 (23%) 8/32 (25%) 13/170 (8%) Preserved RV function 1069/1374 (78%)b 78/92 (85%)b 503/537 (94%) Dilated RV 336/1438 (23%)b 27/103 (26%)b 45/551 (8%) RV hypertrophy 58/665 (9%) 1/45 (2%) 16/266 (6%) RWMA 155/665 (23%)a,b 5/56 (9%) 47/362 (13%)

Percentages excludemissing values.The Bonferroni correction forpost hoc tests was applied.N, number of patients;CAD, coronary artery disease; ACEI, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker; Creat, creatinine; UACR, urine albumin-to- creatinine ratio; n,numberof patientswith available data; Hb,hemoglobin;HbA1c,glycated hemoglobin; BNP, brainnatriureticpeptide; EF, ejection fraction; LVEF, left ventricular ejection fraction; LVd, left ventricular diastolic diameter; LVMi, left ventricular mass index (corrected for body surface area); LAd, left atrium diameter; PAP, pulmonary arterial pressure; RAP, right atrial pressure; RV, right ventricular; RWMA, regional wall motion abnormalities. aP,0.05 versus cardiorenal type 2. bP,0.05 versus cardiorenal type 4. Clin J Am Soc Nephrol 12: 1624–1633, October, 2017 Epidemiology and Natural History of Cardiorenal Syndromes, Mavrakanas et al. 1629

Table 3. Cox proportional hazards regression model for the association between baseline characteristics and the cardiorenal syndrome

HR (95% CI) for Patients HR (95% CI) Baseline Model P Value with Baseline P Value for All Patients CHF and/or CKD

Age per 1 yr 1.01 (1.00 to 1.01) ,0.001 1.00 (0.99 to 1.00) 0.97 Men 1.13 (1.04 to 1.22) 0.004 1.04 (0.95 to 1.13) 0.39 Black race 1.41 (1.25 to 1.59) ,0.001 1.30 (1.15 to 1.48) ,0.001 Hypertension 0.85 (0.78 to 0.93) ,0.001 0.80 (0.73 to 0.87) ,0.001 Diabetes 1.29 (1.18 to 1.41) ,0.001 1.27 (1.16 to 1.39) ,0.001 Coronary disease 1.08 (0.99 to 1.18) 0.07 1.10 (1.00 to 1.21) 0.04 Heart failure 2.12 (1.95 to 2.30) ,0.001 1.14 (1.04 to 1.25) ,0.01 eGFR, per 10 ml/min per 1.73 m2 0.73 (0.72 to 0.75) ,0.001 0.83 (0.81 to 0.84) ,0.001 Baseline model plus LVEF Group 4 ($55%) Reference Reference Group 3 (40%–54%) 1.41 (1.26 to 1.58) ,0.001 1.30 (1.15 to 1.46) ,0.001 Group 2 (25%–39%) 1.84 (1.61 to 2.10) ,0.001 1.66 (1.45 to 1.90) ,0.001 Group 1 (,25%) 2.51 (2.15 to 2.93) ,0.001 2.24 (1.91 to 2.61) ,0.001 Baseline model plus LVEF plus PAP PAP quartile 1 Reference Reference PAP quartile 2 1.07 (0.90 to 1.26) 0.44 0.97 (0.81 to 1.16) 0.75 PAP quartile 3 1.41 (1.20 to 1.65) ,0.001 1.27 (1.07 to 1.51) ,0.01 PAP quartile 4 1.93 (1.66 to 2.25) ,0.001 1.67 (1.42 to 1.95) ,0.001 Baseline plus LVEF plus PAP plus RV Normal RV diameter Reference Reference Dilated RV 1.41 (1.23 to 1.60) ,0.001 1.47 (1.28 to 1.69) ,0.001

HR, hazard ratio; 95% CI, 95% confidence interval; CHF, congestive heart failure; LVEF, left ventricular ejection fraction; PAP, pul- monary arterial pressure; RV, right ventricle.

Figure 4 shows the Kaplan–Meier curves for nonfatal MI associated with mortality, the acute form having the worst and stroke of the different groups. The event distributions prognosis. Both forms were also independently associated were significantly different. Patients with the acute form with MI and stroke occurrence. and type 4 had higher nonfatal MI or stroke rates compared Patients with baseline CKD who progress to HF (type 4) with patients with stable CKD, with the acute form having may also present later on with an acute cardiorenal the highest HR for nonfatal MI or stroke (Table 4). syndrome (Figure 2). It is well established that CKD can Among patients with acute cardiorenal syndrome, we be associated with higher incidence of cardiovascular identified two distinct subgroups: those with or without events, including HF hospitalizations (12). Furthermore, HF and CKD at baseline. Despite significant differences for AKI is common in patients with CKD (16). These associ- most baseline characteristics, the two subgroups had ations may explain the relatively high incidence of admis- similar mortality, MI, and stroke rates. sion with an acute cardiorenal syndrome among patients initially classified in the chronic cardiorenal syndrome type 4group. Discussion Several clinical parameters have been identified to be In this retrospective cohort study of patients referred for associated with worsening kidney function among patients echocardiography, we found that patients who develop admitted for HF, including sex, history of hypertension, cardiorenal syndrome have more underlying comorbidities preexisting HF or atrial fibrillation, diabetes mellitus, heart and worse echocardiographic parameters compared with rate, volume status, baseline creatinine, and systolic BP patients with HF and stable kidney function or patients (17–20). In this study, along with the baseline clinical with CKD and stable heart function. Our data also suggest characteristics, we identified three echocardiographic pa- that the acute and chronic categories may have utility in rameters, LVEF, pulmonary artery pressure, and RV di- identifying distinct patient populations. Patients with an ameter, that were independently associated with the acute form of cardiorenal syndrome had more comorbid- occurrence of cardiorenal syndrome. Metra et al. (21) ities and worse echocardiographic parameters compared have reported that LVEF is an independent predictor of with those with the chronic cardiorenal syndrome (type 2 worsening kidney function among patients admitted for or type 4). However, there was change over the follow-up acute HF. On the contrary, in the Evaluation Study of time, because 19% of patients with a chronic form sub- Congestive Heart Failure and Pulmonary Artery Catheter- sequently developed acute cardiorenal syndrome, and the ization Effectiveness Trial, hemodynamic parameters, in- inverse occurred in 14% of patients in the acute category. cluding wedge pressure and cardiac index, in patients with Lastly, echocardiographic characteristics were independent pulmonary artery catheter–guided therapy were not in- predictors of the cardiorenal syndrome. Development of dependent predictors of worsening kidney function (22). acute or cardiorenal syndrome type 4 was independently Discrepancy with these findings may be due to the broader 1630 Clinical Journal of the American Society of Nephrology

Figure 3. | Patientswithacuteorcardiorenalsyndrometype4have higher mortalityrates comparedwith patientswithheartfailureatbaseline and stable kidney function or with patients with CKD at baseline and stable heart function. Log rank tests: overall: P,0.001; acute cardiorenal syndrome (type 1 or 3; acute) versus control patients with heart failure at baseline and stable kidney function (control 1): P,0.001; acute versus control patients with CKD at baseline and stable heart function (second control group; control 2): P,0.001; type 4 cardiorenal syndrome (CKD at baseline with worsening heart function; type 4) versus control 1: P,0.001; type 4 versus control 2: P,0.001; acute versus type 4: P=0.001. Type 2, type 2 cardiorenal syndrome (heart failure at baseline with worsening renal function). definition of the cardiorenal syndrome in our report, which patients with CKD without any history of hospital admis- also included patients with the chronic forms of the syndrome. sion or outpatient consultation for HF in attempt to clearly We also examined mortality and MI and stroke rates in describe the natural history of the chronic forms and the patients with different types of the cardiorenal syndrome. relative effect of HF and CKD on outcomes. For the same Although many studies have already evaluated clinical reason, we did not assess patients with both CKD and HF outcomes in this population, the coexistence of CKD and at baseline for kidney outcomes (doubling of serum HF in many of them was a major confounder (1,23). creatinine or incidence of ESRD) or HF admissions but Furthermore, no direct comparison between the different only examined for acute forms of the syndrome in this forms was conducted. We included 5107 patients with HF subgroup. We found that patients with the acute form have and a baseline eGFR $60 ml/min per 1.73 m2 and 4129 the worst prognosis followed by those with cardiorenal

Table 4. Cox proportional hazards regression model for the association between the various forms of the cardiorenal syndrome and all- cause mortality, nonfatal myocardial infarction, or stroke

Mortality Nonfatal MI Stroke Group HR (95% CI) P Value HR (95% CI) P Value HR (95% CI) P Value

Control group 2 Reference Reference Reference (stable CKD) Control group 1 1.37 (1.25 to 1.50) ,0.001 1.18 (1.01 to 1.39) 0.04 0.93 (0.74 to 1.17) 0.53 (stable HF) Acute 3.13 (2.72 to 3.61) ,0.001 3.36 (2.41 to 4.69) ,0.001 1.98 (1.35 to 2.91) 0.001 Type 2 2.41 (0.90 to 6.44) 0.08 Type 4 1.49 (1.20 to 1.85) ,0.001 2.21 (1.42 to 3.45) ,0.001 1.76 (1.05 to 2.95) 0.03 Acute versus type 2 1.30 (0.48 to 3.50) 0.60 Type 4 versus type 2 0.62 (0.23 to 1.69) 0.35 Acute versus type 4 2.10 (1.65 to 2.68) ,0.001 1.52 (0.89 to 2.59) 0.12 1.12 (0.61 to 2.06) 0.71

No myocardial infarctions or strokes were observed in patients with chronic cardiorenal (type 2) syndrome after the landmark date. HRs were adjusted for age, sex, race, and history of hypertension, diabetes, and coronary artery disease. HR, hazard ratio; 95% CI, 95% confidence interval; HF, heart failure. Clin J Am Soc Nephrol 12: 1624–1633, October, 2017 Epidemiology and Natural History of Cardiorenal Syndromes, Mavrakanas et al. 1631

Figure 4. | Patients with acute or cardiorenal syndrome type 4 have higher (A) myocardial infarction (MI) or (B) stroke rates compared with patients with heart failure at baseline and stable kidney function or with patients with CKD at baseline and stable heart function. No MIs or strokes were observed in patients with chronic cardiorenal syndrome (type 2) after the landmark date. Log rank tests for nonfatal MI: overall: P,0.001; acute cardiorenal syndrome (type 1 or 3; acute) versus control patientswith heart failure at baseline and stable kidney function (control 1): P,0.001; acute versus control patients with CKD at baseline and stable heart function (second control group; control 2): P,0.001; type 4 cardiorenal syndrome (CKD at baseline with worsening heart function; type 4) versus control 1: P,0.01; type 4 versus control 2: P,0.001; acute versus type 4: P=0.11. Log rank tests for stroke: overall: P,0.001; acute versus control 1: P,0.001; acute versus control 2: P,0.001; type 4 versus control 1: P=0.001; type 4 versus control 2: P=0.03; acute versus type 4: P=0.78. syndrome type 4. One may argue that the former condition Our study has several limitations. It is a retrospective reflects underlying heart disease, whereas the latter may observational study of patients referred for echocardio- primarily reflect volume retention in patients with slowly gram that may not represent the full spectrum of patients declining kidney function. Too few patients with cardiore- with HF and/or CKD. We did not have specificinforma- nal syndrome type 2 were identified in the landmark tion on the indications for echocardiography, which were analysis to draw definitive conclusions. interpreted by different clinical cardiologists and were not 1632 Clinical Journal of the American Society of Nephrology

interpreted by a central standard laboratory. Thus, our 2. Ronco C, Haapio M, House AA, Anavekar N, Bellomo R: Car- echocardiographic data, although largely accurate, may diorenal syndrome. J Am Coll Cardiol 52: 1527–1539, 2008 3. Ronco C, McCullough P, Anker SD, Anand I, Aspromonte N, lack the precision of research quantitative techniques, albeit Bagshaw SM, Bellomo R, Berl T, Bobek I, Cruz DN, Daliento L, nondirectionally biased. Comorbidities and clinical out- Davenport A, Haapio M, Hillege H, House AA, Katz N, Maisel A, comes were identified on the basis of diagnostic codes. Mankad S, Zanco P, Mebazaa A, Palazzuoli A, Ronco F, Shaw A, Billing codes have low sensitivity in identifying AKI Sheinfeld G, Soni S, Vescovo G, Zamperetti N, Ponikowski P; episodes, especially less severe ones (24,25). The main Acute Dialysis Quality Initiative (ADQI) consensus group: Cardio- renal syndromes: Report from the consensus conference of analysis excluded an important number of events that the acute dialysis quality initiative. Eur Heart J 31: 703–711, occurred before the landmark date, which was arbitrarily 2010 selected. Additionally, we combined patients with acute 4. McCullough PA, Kellum JA, Haase M, Mu¨ller C, Damman K, forms (types 1 and 3) into a single category, and we did not Murray PT, Cruz D, House AA, Schmidt-Ott KM, Vescovo G, Bagshaw SM, Hoste EA, Briguori C, Braam B, Chawla LS, account for type 5 (systemic disorders that concurrently Costanzo MR, Tumlin JA, Herzog CA, Mehta RL, Rabb H, Shaw influence cardiac and kidney function), because this could AD, Singbartl K, Ronco C: Pathophysiology of the cardiorenal not be accurately performed in a retrospective study. For syndromes: Executive summary from the eleventh consensus patients with acute forms, the temporal sequence of events conference of the Acute Dialysis Quality Initiative (ADQI). Contrib Nephrol 182: 82–98, 2013 could not be ascertained, and patients with type 5 may have 5. Haase M, Mu¨ller C, Damman K, Murray PT, Kellum JA, Ronco C, been classified as acute cardiorenal. Nevertheless, the McCullough PA: Pathogenesis of cardiorenal syndrome type 1 in sample size of this study, the assessment of multiple acute decompensated heart failure: Workgroup statements echocardiographic parameters, a sufficient number of out- from the eleventh consensus conference of the Acute Dialysis Quality Initiative (ADQI). Contrib Nephrol 182: 99–116, 2013 comes in most groups, and most importantly, the discrim- 6. Damman K, Valente MA, Voors AA, O’Connor CM, van ination between the two main disease processes at baseline Veldhuisen DJ, Hillege HL: Renal impairment, worsening renal (HF and CKD) represent unique strengths of our analysis. function, and outcome in patients with heart failure: An updated Diagnostic codes for HF, when used in administrative meta-analysis. Eur Heart J 35: 455–469, 2014 datasets, successfully reflect similar trends in epidemiologic 7. Testani JM, Chen J, McCauley BD, Kimmel SE, Shannon RP: Po- tential effects of aggressive decongestion during the treatment of studies (26). Many of the results reported in this study might decompensated heart failure on renal function and survival. have been anticipated on the basis of current knowledge but Circulation 122: 265–272, 2010 have never been specifically analyzed to assess whether the 8. Bagshaw SM, Hoste EA, Braam B, Briguori C, Kellum JA, various cardiorenal syndrome categories have clinical utility. McCullough PA, Ronco C: Cardiorenal syndrome type 3: Pathophysiologic and epidemiologic considerations. Contrib Thus, our study provides important new information com- Nephrol 182: 137–157, 2013 pared with the existing literature, which had not previously 9. Elsayed EF, Tighiouart H, Griffith J, Kurth T, Levey AS, Salem D, identified clinical distinctions or differences in prognosis of Sarnak MJ, Weiner DE: Cardiovascular disease and subsequent different cardiorenal syndrome classes. In conclusion, pa- kidney disease. Arch Intern Med 167: 1130–1136, 2007 10. Cruz DN, Schmidt-Ott KM, Vescovo G, House AA, Kellum JA, tients with the acute and chronic forms of the cardiorenal Ronco C, McCullough PA: Pathophysiology of cardiorenal syn- syndrome have different baseline characteristics and echo- drome type 2 in stable chronic heart failure: Workgroup state- cardiogram parameters. Some patients with a chronic form ments from the eleventh consensus conference of the Acute will subsequently develop an acute cardiorenal syndrome, Dialysis Quality Initiative (ADQI). Contrib Nephrol 182: 117– 136, 2013 and some patients with acute cardiorenal syndrome will 11. Foley RN, Murray AM, Li S, Herzog CA, McBean AM, Eggers PW, progress to CKD or HF. The echocardiogram characteristics Collins AJ: and the risk for cardiovascular are independent predictors of the cardiorenal syndrome. disease, renal replacement, and death in the United States Development of the acute form or cardiorenal syndrome type medicare population, 1998 to 1999. J Am Soc Nephrol 16: 489– 495, 2005 4 is independently associated with mortality, the acute form 12. Go AS, Chertow GM, Fan D, McCulloch CE, Hsu CY: Chronic having the worst prognosis. kidney disease and the risks of death, cardiovascular events, and hospitalization. N Engl J Med 351: 1296–1305, 2004 Acknowledgments 13. Tumlin JA, Costanzo MR, Chawla LS, Herzog CA, Kellum JA, T.A.M. was supported by a grant from the Foundation G. and McCullough PA, Ronco C: Cardiorenal syndrome type 4: Insights on clinical presentation and pathophysiology from the eleventh L. Hirsch (General Internal Medicine Division, Geneva University consensus conference of the Acute Dialysis Quality Initiative Hospitals). A.K. and K.S. were supported by the National Institute (ADQI). Contrib Nephrol 182: 158–173, 2013 of Diabetes and Digestive and Kidney Diseases training grant 14. Mehta RL, Rabb H, Shaw AD, Singbartl K, Ronco C, McCullough T32DK007527. PA, Kellum JA: Cardiorenal syndrome type 5: Clinical pre- sentation, pathophysiology and management strategies from the Disclosures eleventh consensus conference of the Acute Dialysis Quality Initiative (ADQI). Contrib Nephrol 182: 174–194, 2013 None. 15. Levey AS, Stevens LA, Schmid CH, Zhang YL, Castro AF 3rd, Feldman HI, Kusek JW, Eggers P, Van Lente F, Greene T, Coresh J; CKD-EPI (Chronic Kidney Disease Epidemiology Collaboration): References A new equation to estimate glomerular filtration rate. Ann Intern 1. Bagshaw SM, Cruz DN, Aspromonte N, Daliento L, Ronco F, Med 150: 604–612, 2009 Sheinfeld G, Anker SD, Anand I, Bellomo R, Berl T, Bobek I, 16. Hsu CY,Ordonez~ JD, Chertow GM, Fan D, McCulloch CE, Go AS: Davenport A, Haapio M, Hillege H, House A, Katz N, Maisel A, The risk of acute renal failure in patients with chronic kidney Mankad S, McCullough P, Mebazaa A, Palazzuoli A, Ponikowski disease. Kidney Int 74: 101–107, 2008 P, Shaw A, Soni S, Vescovo G, Zamperetti N, Zanco P, Ronco C; 17. Forman DE, Butler J, Wang Y, Abraham WT, O’Connor CM, Acute Dialysis Quality InitiativeConsensus Group: Epidemiology Gottlieb SS, Loh E, Massie BM, Rich MW, Stevenson LW,Young JB, of cardio-renal syndromes: Workgroup statements from the 7th Krumholz HM: Incidence, predictors at admission, and impact of ADQI consensus conference. Nephrol Dial Transplant 25: 1406– worsening renal function among patients hospitalized with heart 1416, 2010 failure. J Am Coll Cardiol 43: 61–67, 2004 Clin J Am Soc Nephrol 12: 1624–1633, October, 2017 Epidemiology and Natural History of Cardiorenal Syndromes, Mavrakanas et al. 1633

18. Krumholz HM, Chen YT, Vaccarino V, Wang Y, Radford MJ, heart failure: Systematic review and meta-analysis. J Am Coll Bradford WD, Horwitz RI: Correlates and impact on outcomes of Cardiol 47: 1987–1996, 2006 worsening renal function in patients . or =65 years of age with 24. Grams ME, Waikar SS, MacMahon B, Whelton S, Ballew SH, heart failure. Am J Cardiol 85: 1110–1113, 2000 Coresh J: Performance and limitations of administrative data in 19. Cowie MR, Komajda M, Murray-Thomas T,Underwood J, Ticho B; the identification of AKI. Clin J Am Soc Nephrol 9: 682–689, POSH Investigators: Prevalence and impact of worsening renal 2014 function in patients hospitalized with decompensated heart 25. Waikar SS, Wald R, Chertow GM, Curhan GC, Winkelmayer WC, failure: Results of the prospective outcomes study in heart failure Liangos O, Sosa MA, Jaber BL: Validity of international classifi- (POSH). Eur Heart J 27: 1216–1222, 2006 cation of diseases, ninth revision, clinical modification codes for 20. Butler J, Forman DE, Abraham WT, Gottlieb SS, Loh E, Massie BM, acute renal failure. J Am Soc Nephrol 17: 1688–1694, 2006 O’Connor CM, Rich MW, Stevenson LW, Wang Y, Young JB, 26. McCullough PA, Philbin EF, Spertus JA, Kaatz S, Sandberg KR, Krumholz HM: Relationship between heart failure treatment and Weaver WD; Resource Utilization Among Congestive Heart development of worsening renal function among hospitalized Failure (REACH) Study: Confirmation of a heart failure epidemic: patients. Am Heart J 147: 331–338, 2004 Findings from the Resource Utilization Among Congestive Heart 21. Metra M, Nodari S, Parrinello G, Bordonali T, Bugatti S, Danesi R, Failure (REACH) study. J Am Coll Cardiol 39: 60–69, 2002 Fontanella B, Lombardi C, Milani P, Verzura G, Cotter G, Dittrich H, Massie BM, Dei Cas L: Worsening renal function in patients Received: April 11, 2017 Accepted: June 29, 2017 hospitalised for acute heart failure: Clinical implications and prognostic significance. Eur J Heart Fail 10: 188–195, 2008 22. Nohria A, Hasselblad V,Stebbins A, Pauly DF, Fonarow GC, Shah Published online ahead of print. Publication date available at www. M, Yancy CW, Califf RM, Stevenson LW, Hill JA: Cardiorenal cjasn.org. interactions: Insights from the ESCAPE trial. J Am Coll Cardiol 51: 1268–1274, 2008 This article contains supplemental material online at http://cjasn. 23. Smith GL, Lichtman JH, Bracken MB, Shlipak MG, Phillips CO, asnjournals.org/lookup/suppl/doi:10.2215/CJN.04020417/-/ DiCapua P, Krumholz HM: Renal impairment and outcomes in DCSupplemental.