suPAR is associated with risk of future acute surgery and post-operative mortality in acutely admitted medical patients

This retrospective, registry-based study was carried out at Copenhagen University Hospital Amager and Hvidovre, Hvidovre, Denmark, investigating patients who were admitted and had blood samples drawn at the Acute Medical Unit (AMU) from 18 November 2013 until 30 September 2015. Since 18 November 2013, suPAR was routinely analyzed in all AMU patients as part of the standard blood samples taken at admission. Data on a subgroup of this population was previously described by Rasmussen et al. [1]. The index admission was defined as the first admission during this period where suPAR was measured. The patients were followed in national registries for minimum 90 days from the index admission through 31 December 2015. The Danish personal civil registration number (CPR number) was used to link patient data from national registries with laboratory data from the electronic hospital database LABKA. The Danish National Patient Registry (NPR) contains information on all hospital admissions, including dates of admission and discharge, registered diagnoses, and surgical procedures performed. Hospital admission contacts in the NPR with less than 5 h apart were considered to belong to the same hospital admission, while contacts with more than 5 h apart were considered to belong to separate admissions. The survival status was extracted from the Civil Registration System.

Patients were included if they had standard blood samples analyzed during their admission to the AMU. Children, obstetrical-, gynecological-, gastroenterological-, and acute surgical patients were admitted to specialized acute departments and were therefore not included. Patients were excluded if no suPAR measurement was available or if follow-up via national registries was not possible.

The primary outcome was surgery, defined as a patient having any surgical procedure registered in the NPR during an admission in the follow-up period: acute surgery was defined as an admission containing at least one acute surgical procedure; elective surgery was defined as an admission containing only planned, and no acute, surgical procedures. Thus, if a patient had both an elective and an acute surgical procedure registered within the same admission, it was defined as acute surgery. The distinction between acute and elective surgical procedures was made based on a variable in NPR that indicates whether the admission for a given surgical contact is acute or planned. Surgical procedures were stratified according to specialty, using the classification from the Danish national Healthcare Classification System (SKS). Smaller surgical procedures, such as lumbar- and ascites puncture and needle biopsies, and investigative procedures, such as endoscopies, were included in the analyses as surgery. A separate analysis excluding patients who only had minor surgery or endoscopies was carried out as well. Surgery types with less than 50 cases were compiled into the category “Other” (endocrine-, ophthalmological-, head and neck-, oral-, breast-, and tissue transplantation surgical procedures). Cancer-related surgery was defined as any surgical procedure for the primary diagnosis cancer (ICD-10 C00-C97). It should be noted that a patient can be in more than one surgical specialty.

The secondary outcome was all-cause mortality. When comparing mortality between non-surgical and surgical patients, we used mortality rates per person-year during the entire follow-up period. Patients who received surgery were categorized as non-surgical patients until the day of surgery; after that their time at risk was calculated for acute or elective surgery, respectively, until death or end of follow-up. This was done to prevent immortality bias. When comparing mortality between surgical specialties, we used 90-day post-operative mortality calculated from the first day of surgery.

Plasma levels of suPAR were analyzed as part of the standard blood samples at the Department of Clinical Biochemistry as previously described [1]. Briefly, suPAR was measured daily during weekdays in singlets using the suPARnostic® AUTO Flex ELISA on a Siemens BEP2000 analyzer according to the manufacturer’s instructions (ViroGates A/S, Birkerød, Denmark). C-reactive protein (CRP) was measured using a COBAS 6000 analyzer (Roche Diagnostics, Mannheim, Germany).

The Charlson Comorbidity Index (Charlson score) [21], was calculated using all diagnoses registered in the NPR for the index admission and in the 2 years prior to the index admission. The Charlson Score is a weighted scoring system, rating the number and severity of chronic comorbidities, including chronic kidney disease among others. We used a SAS macro (developed by Ken Turner and Charles Burchill) based on ICD-10 diagnoses to calculate the Charlson score [22], using the revised weighting by Quan et al. [23].

The risk of acute surgery and post-operative mortality was modeled by Cox regression as a function of log2-transformed suPAR adjusted for sex, age, CRP, and Charlson score. A second analysis was made including creatinine in the adjustment. We used a combined endpoint for acute surgery, using death and elective surgery as competing risks. In the Cox regression analysis of post-operative mortality, the number of days to surgery was also included as an explanatory variable to adjust for differences in time from suPAR measurement to surgery. We also tested for an interaction between the suPAR level and time from suPAR analysis to surgery. The results are presented as hazard ratios (HRs) with 95% confidence intervals (CIs).

The risk of acute surgery and 90-day post-operative mortality according to suPAR level was presented in cumulative incidence plots and Kaplan-Meier plots, respectively. suPAR values were stratified in 3 ng/ml intervals to ensure that the results were less dependent on the specific cohort and thereby comparable with other patient cohorts. The log-rank test was used to compare mortality or risk of acute surgery between suPAR strata in the cumulative incidence and Kaplan-Meier plots.

The cohort consisted of 20,193 patients who were admitted to the AMU for an acute medical condition. A total of 2881 were excluded (Fig. 1), leaving 17,312 patients for analysis. Baseline characteristics are shown in Table 1. The median follow-up time for all included patients was 394 days (IQR 218–583). During follow-up, 3768 (21.8%) patients had surgical procedures performed. Of these, 2404 (63.8%) were acute (Table 2). The median time from suPAR measurement at the index admission until surgery was 74 days (IQR 9–218); 45 days (IQR 5–186) for acute surgery and 118 days (IQR 37–275) for elective surgery.

Patients receiving acute surgery were older and had higher suPAR and CRP levels at the index admission compared with non-surgical patients (Table 2). Furthermore, the median Charlson score was higher, showing an increased prevalence of comorbidities among acute surgical patients (Table 2). Similarly, patients who underwent elective surgery were older and had higher suPAR and Charlson score compared with non-surgical patients (Table 2). For the entire population, suPAR and CRP were positively correlated (Kendall’s tau b 0.36, p < 0.0001).

In the Cox regression model adjusted for sex, age, CRP, and Charlson score, suPAR was associated with acute surgery during follow-up, with a HR of 1.50 (95% CI: 1.42–1.59, p < 0.0001) for every doubling in suPAR. When including creatinine in the adjustment, the HR was 1.45 (95% CI: 1.37–1.54, p < 0.0001).

Furthermore, we carried out a sub-analysis excluding 761 patients that exclusively had minor surgical procedures performed (minor surgery and endoscopies). In this sub-analysis, adjusted for sex, age, CRP and Charlson score, every doubling in suPAR was associated with a HR of 1.37 (95% CI: 1.29–1.47, p < 0.0001). Finally, the cumulative incidence rate of acute surgery increased with increasing suPAR (Fig. 2).

When comparing acute with elective surgery, suPAR was increased in patients undergoing endoscopies, gastroenterological-, orthopedic-, respiratory-, urological- and, to a lesser degree, dermatological- and cancer-related surgical procedures (Table 3).

In the entire population of acutely admitted medical patients, mortality rates were higher among those who received acute surgery during follow-up compared with non-surgical patients, whereas mortality rates were similar among patients receiving elective surgery compared with the non-surgical patients (Table 2).

90-day post-operative mortality was higher in patients receiving acute surgery compared to elective in the majority of specialties (Table 3). This increased mortality was particularly evident for gastroenterological-, orthopedic-, respiratory-, vascular-, and cancer-related surgery as well as endoscopies, but also observed for cardiothoracic-, neurological-, urological-, minor-, and dermatological surgical procedures (Table 3).

Of the patients receiving acute surgery, 439 died (18.3%) within 90 days follow-up after surgery. In Cox regression analysis adjusted for sex, age, CRP, and Charlson score, the HR for 90-day post-operative mortality for doubling the suPAR value was 1.73 (95% CI: 1.52–1.95, p < 0.0001). Due to variability in time from the index admission to the acute surgical procedure, we tested for an interaction between the suPAR level and time from suPAR analysis to surgery in the Cox regression analysis for 90-day post-operative mortality, and found no interaction (p = 0.86). The 90-day post-operative mortality after acute surgery was higher for patients with elevated baseline levels of suPAR (Fig. 3).