High sensitivity troponin-I and myocardial injury after noncardiac surgery for cancer in a large prospective cohort Authors: Todd Liu, Patrick J McCormick, David Amar, Kay See Tan, Jasme Lee, Gregory W Fischer, Laurel Jackson, David Daghfal, Lakshmi V Ramanathan, Carol L Chen, Gillian Murtagh, Amanda G Blouin Published: 2025 ABSTRACT AbstractBackgroundMyocardial injury after noncardiac surgery (MINS) impact in cancer is unclear. In a general population, high-sensitivity troponin-I assays (hsTnI) have better MINS detection than contemporary troponin-I (TnI) assays. This study summarizes MINS prevalence in cancer surgery patients using hsTnI and TnI. The primary endpoint was 30-day mortality. Secondary endpoints included 90-day mortality, and observed-to-expected length of stay ratio [O: E LOS].MethodsMINS was explored using three postoperative elevation definitions: MINSTnI: TnI ≥ 0.06 ng/mL, MINShsTnI: hsTnI ≥ 17 ng/L (female) ≥ 35 ng/L (male), or MINSeither: MINSTnI or MINShsTnI. MINS and patient characteristics were analyzed for univariable associations with 30-day mortality using logistic regression. The association between MINS and cardiotoxic cancer therapy was assessed with Chi-squared test, and association with secondary endpoints were quantified with regression analyses.ResultsAmong 4,278 patients, prevalence of MINSTnI, MINShsTnI, and MINSeither were 4.6%, 9.3%, and 9.5%. In univariable analysis, MINShsTnI was associated with greater odds of 30-day mortality (OR 3.3; 95% CI 1.07, 8.57, p = 0.039), but association with MINSTnI was not detected (OR 3.67; 95% CI 0.85, 11.0, p = 0.076). In multivariable Cox analysis, association between MINShsTnI and 90-day mortality was not detected (HR 1.46, 95% CI 0.78, 2.71, p = 0.2). Cardiotoxic cancer therapy had increased odds of 30-day mortality in univariable analysis (OR 3.20; 95% CI 1.25, 7.75, p = 0.017) and higher MINShsTnI prevalence versus non-exposed patients (12% vs. 9%, p = 0.015). Patients with MINShsTnI had a 25% (95% CI 16%, 32%, p < 0.001) higher O: E LOS ratio.ConclusionThis study of MINS in cancer surgery observed increased odds of 30-day mortality in univariable analysis. An association with 90-day mortality, however, was not detected in multivariable analysis. MINS was associated with an increased O: E LOS ratio. As expected, the hsTnI assay detected more MINS events. The relationship of cardiotoxic cancer therapies to MINS risk warrants investigation.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12871-025-03488-y.Keywords: Myocardial injury after noncardiac surgery (MINS), High-Sensitivity Troponin, Cancer Surgery Outcomes, 30-Day Mortality, Troponin, Cardiotoxic chemotherapy, Postoperative cardiac complications, Length of Stay, Perioperative Risk Stratification Introduction Up to 42% of all perioperative mortality during non-cardiac surgery has been associated with myocardial injury [1, 2], defined as elevated cardiac troponin values above the 99th percentile upper reference limit [3]. Recent improvements in the analytic performance of Tn assays may be expected to result in improved ability to detect and monitor myocardial injury in the setting of non-cardiac surgery. Myocardial injury after non-cardiac surgery (MINS) is defined as myocardial injury caused by ischemia that occurs during or within 30 days of surgery. MINS was first comprehensively investigated by the Vascular Events In Noncardiac Surgery Patients Cohort Evaluation (VISION) study which found that elevated troponin after noncardiac surgery independently predicted 30-day mortality. A subsequent VISION study found that an elevated postoperative high sensitivity troponin-T level (hsTnT) during the first 3 days after surgery was significantly associated with 30-day mortality [2, 4, 5]. In nonsurgical populations, hsTnI is more strongly associated with cardiovascular outcomes than TnT [6]. The prevalence of MINS was originally reported as 8% of all noncardiac surgery patients [4, 5]. However, 84% of patients with MINS have no symptoms [7, 8]. Without postoperative troponin monitoring, MINS in these patients would not have been detected. Cardiac risk scoring algorithms may not adequately account for the risk factors for cancer surgery patients from their oncologic diagnosis and/or associated therapies [9]. The inflammatory pathophysiology of cancer is linked to cardiovascular risk factors and therefore injury [10]. Anemia, physical deconditioning, cancer-related organ dysfunction, and metabolic imbalances are prevalent in cancer surgery patients. Cancer therapies such as anthracyclines, trastuzumab, tyrosine kinase inhibitors, VEGF inhibitors, and immune checkpoint inhibitors can also increase the risk of major adverse cardiac events [11, 12]. While the VISION study used high sensitivity troponin (hsTn) testing, many hospitals are still using contemporary troponin testing. High sensitivity troponin assays can detect measurable troponin in >50% of a general reference population with coefficients of variation < 10% at and below the 99th percentile. HsTn assays are now preferred over contemporary assays in the setting of suspected myocardial infarction (MI) [13], but have not yet been comprehensively evaluated in a large cohort of cancer patients undergoing surgery. The aim of this study is to quantify the association of 30-day mortality after cancer surgery with MINS as detected by TnI and hsTnI assays. We hypothesized that hsTnI assay will detect a larger proportion of MINS patients than the contemporary TnI assay in this cancer population. This study also assessed the relationship between MINS and 90-day mortality rates, observed to expected length of stay ratio, and postoperative complications. Methods Ethics statements The study was approved by the Institutional Review Board (IRB) of Memorial Sloan Kettering Cancer Center (approval IRB# 17–643). The study protocol included a request for waiver of informed consent that was approved by the institutional review board as the study protocol involved minimal risk to participants or their privacy, and that the data will be deidentified and destroyed upon completion of the project. Study design and participants This prospective cohort study included adults (≥ 18 years old) who underwent elective major noncardiac cancer surgery at our institution under general anesthesia requiring at least one night stay in the hospital. All blood specimens were obtained from waste blood as part of routine care for patients. Study related troponin results were not reported to the electronic medical record or practitioners and therefore did not alter the care each patient received. Excluded from the study were patients with known nonischemic etiologies for troponin elevation (i.e. sepsis, end-stage renal disease) and those with elevated preoperative troponin levels. Also excluded were patients undergoing breast surgery because of the low frequency of required perioperative blood sampling. Troponin assay procedure Patients meeting eligibility criteria were screened for routine preoperative testing ordered on EDTA blood tubes drawn within 30 days of the scheduled procedure. If present, these samples were collected and stored separately. Following the surgery, EDTA plasma samples were collected and stored for postoperative day (POD) zero, one, two, and three when available. Spun plasma was stored at −80 degrees Celsius. ​After thawing, plasma was centrifuged prior to analysis (3,400 RCF, 30 min, room temperature)​​. Troponin-I was measured using both the contemporary Troponin-I and the high sensitivity Troponin-I assays on an Abbott ARCHITECT i2000 analyzer (Abbott Laboratories, Abbott Park, IL).​ All samples were analyzed using both assays unless recovered volumes were inadequate. Assay performance was verified by the clinical laboratory. For hsTnI, linearity was confirmed from 0.5 ng/L to 50,000 ng/L; precision studies yielded CVs < 10%, and limit of quantitation (with a CV < 10%) was determined to be 4.6ng/L. For TnI, the verified linear range was 0.01 to 50 ng/mL (equal to 10–50,000 ng/L); precision at 0.25 ng/mL, was 4.79% CV. The verified analytical sensitivity was 0.01 ng/mL (equal to 10 ng/L). Comparison studies between the hsTnI and TnI assays on patient samples demonstrated good correlation. Laboratory evaluation Three categories defined MINS, based on the troponin-I assay used: [1] MINSTnI [2] MINShsTnI [3] MINSeither. Patients with a post-operative (any POD 0–3) troponin-I level ≥ 0.06 ng/mL (equal to 60 ng/L) as measured by the contemporary assay are categorized as MINSTnI. This cutoff was based on our local clinical threshold at the time of study initiation. MINShsTnI was defined as any postoperative (any POD 0–3) high-sensitivity troponin-I ≥ 17ng/L for females and ≥ 35ng/L for males. This cutoff was based on the 99th percentile value from assay manufacturer. MINSeither included patients who either had MINSTnI or MINShsTnI. Sample collection and follow-up The enrollment periods were November 2018 through July 2019 and November 2019 through January 2020. Following each enrollment period, samples were processed and tested. The pause in enrollment was due to exhausted storage capacity for samples. Enrollment resumed as soon as sufficient samples had been tested to allow for storage of new samples. Comorbidities along with outcomes and complications data were obtained from the electronic medical record using ICD-10 diagnosis codes and the cancer registry for death date. Complications were assessed using Elixhauser comorbidity definitions [14]. Expected length of stay values are from the Vizient Clinical Data Base (Vizient, Irving, TX) calculated using the Vizient risk model applied to the patient’s discharge claim data. Exposure to cardiotoxic therapeutics within three months was obtained through the electronic medical record using a list of exposure agents. (Supplementary Table S1) Patient mortality was determined through the institution’s vital status monitoring program which includes family reports, obituary searches, and the national Social Security Death Index. Patients with no clinical notes, laboratories, or other vital status indicator after 30 post-operative days were considered lost to follow-up. Statistical analysis A prespecified analysis plan was determined before specimen collection began. We assumed a surgical inpatient 30-day mortality of 3% for our institution, and previous published studies of MINS found a 2-fold or greater increase in 30-day mortality in patients with an elevated troponin. Our power analysis assumed 6% 30-day mortality in patients with MINSTnI. We assumed a 15% prevalence of MINSTnI, based on the 11.6% prevalence found in the VISION study using contemporary TnT and a 22% prevalence found in a similar study [2, 15]. Our calculated minimum sample size was 3,750 patients with a power of 80%. Patient demographics and troponin measurements were summarized descriptively with median (interquartile range, IQR) for continuous and n (%) for categorical measures. Association between each of the three MINS definitions and binary outcomes (30-day mortality and postoperative complications) were descriptively summarized. MINS and patient characteristics were analyzed for univariable associations with the primary outcome of 30-day mortality using logistic regression. Linear regression was used when analyzing the relationship of the MINS definitions and the log-transformed observed to expected length of stay ratio [O: E LOS]. A Cox proportional hazard model was also used for secondary outcome analysis using 90-day mortality. Variables included in the Cox model were selected based on those used in the VISION study mortality model: active cancer (for which we used recent exposure to cardiotoxic chemotherapy), emergent surgery, history of peripheral vascular disease, history of chronic pulmonary disease, age, neurosurgery, and MINS, with the addition of coronary stent history and congestive heart failure history. Additionally, the association between MINS and cardiotoxic cancer therapy was assessed with Chi-squared test. The investigators performed an exploratory analysis of how alternative troponin thresholds may interact with mortality risk. Previous studies within the asymptomatic general population have categorized patient cohorts into low, intermediate, and high cardiovascular risk (i.e. cardiovascular death, myocardial infarction, heart failure, or stroke) based on hsTnI values [16]. The alternative exploratory thresholds were derived from the ARIC study: for men, < 6 ng/L (low risk), 6 to 12 ng/L (intermediate risk), and >12 ng/L (high risk); for women, < 4 ng/L (low risk), 4 to 10 ng/L (intermediate risk), and >10 ng/L (high risk) [17]. A Cox proportional hazard model was constructed to examine the relationship between these postoperative hsTnI cardiovascular risk categorizes using landmark analysis. Landmark time was defined as 3 days post-surgery. Events were defined as 90-day mortality because 30-day mortality was underpowered given the low event rate. HsTnI cardiovascular risk category was determined by the peak POD hsTnI concentration. Statistical significance was defined by two-sided p-value < 0.05 and all analyses were conducted using R 4.1.3 (R Core Team, 2022). Results The patient flow diagram is presented in (Fig. 1). There were 5923 patient surgeries eligible for the study. After exclusions, 4278 patients were included. Patient follow-up analysis showed that 4241 patients were known to be alive at least until postoperative day 30 and 20 patients (0.5%) were known to have died. Only 17 patients (0.40%) of the study population did not complete 30-day follow up. Based on the low mortality, we assumed these patients were alive for the 30-day mortality endpoint. Patient flow diagram Patient characteristics are described in Table 1. Preoperative samples were available for 69.5% of the population. Postoperative samples were collected on POD 0–3 for 61.3%, 78.8%, 59.8%, and 47.5% of the population, respectively. 49% of the patient population had at least three postoperative blood samples available for both TnI and HsTnI testing. TnI and HsTnI were both measured on each sample concurrently and simultaneously when sample volume permitted. Less than 1% of collected samples were available for only TnI or hsTnI testing, but not both (Supplementary Table S2). Among those with both testing, ten samples had discordant results (TnI was elevated but hsTnI was not elevated), out of 3868 patients (0.2%) with hsTnI not elevated. This finding is consistent with published findings of the rate of discordant results with assays [18, 19]. In practice, clinicians will rarely encounter discordant measurements as usually only one type of troponin assay is available. An elevated TnI without elevated hsTnI is unlikely to be valid and should prompt retesting because the high sensitivity test is more accurate at low concentrations. Table 1Demographics, baseline characteristics, and type of surgeryPatient CharacteristicN = 42781Age (years)61 (50, 70) <45707 (17%) 45 to 641,839 (43%) 65 to 741,147 (27%) >=75585 (14%)Female2246 (53%)BMI (kg/m²)27.1 (23.7, 31.3)Race White3366 (81%) Black302 (7.3%) Asian355 (8.5%) Native American or Alaskan Native6 (0.1%) Native Hawaiian or Other Pacific Islander3 (< 0.1%) Other128 (3.1%) Unknown118Hispanic Ethnicity289 (6.8%)Congestive heart failure130 (3.0%)Valvular disease181 (4.2%)Pulmonary circulation disorders54 (1.3%)Peripheral vascular disease142 (3.3%)Hypertension without end organ complications1638 (38%)Hypertension with end organ complications281 (6.6%)Diabetes without end organ complications410 (9.6%)Diabetes with end organ complications154 (3.6%)Liver disease198 (4.6%)Atrial fibrillation (or flutter)262 (6.1%)Renal failure244 (5.7%)End stage renal disease6 (0.1%)Previous percutaneous coronary intervention165 (3.9%)Exposure to cardiotoxic cancer therapy within three months743 (17%)Surgical Service Colorectal516 (12%) Gastric/Mixed Tumor424 (9.9%) Gynecology504 (12%) Head & Neck242 (5.7%) Hepatopancreatobiliary574 (13%) Neurosurgery378 (8.8%) Orthopedics250 (5.8%) Plastic248 (5.8%) Thoracic572 (13%) Urology570 (13%)Anesthesia Duration (hours)4.68 (3.48, 6.58)Surgical Duration (hours)3.43 (2.25, 5.27)ASA PS2 >= 33354 (78%)General Anesthesia4234 (99%)Emergent Surgery40 (0.9%)Estimated Blood Loss (mL)100 (50, 300)1 Data presented: median (IQR) or n (%) 2ASA PS = American Society of Anesthesiologists Physical Status Classification Score Demographics, baseline characteristics, and type of surgery 1 Data presented: median (IQR) or n (%) 2ASA PS = American Society of Anesthesiologists Physical Status Classification Score MINSTnI and MINShsTnI was found in 198 patients (4.6%) and 395 patients (9.3%), respectively. Taken together, 405 patients (9.5%) had MINSeither. Surgical cases were selected based on the proposed anesthesia type; in 44 cases (1.0%) an anesthesia type other than general was used, and 40 cases (0.9%) were declared to be emergent by the anesthesiologist. Postoperative outcomes All-cause 30-day mortality in the study cohort occurred in 20 patients (0.5%), and 90-day mortality in 88 patients (2.0%). Among the 405 patients with MINSeither, 5 (1.2%) had 30-day mortality and 14 (3.5%) had 90-day mortality. The overall median length of stay among the entire cohort was 4.0 days (IQR 3.0–7.0). Postoperative outcomes are detailed in Table 2. The proportion of postoperative cardiac complications and atrial fibrillation in each MINS categorizations were higher than in non-MINS patients (Supplementary Table S3). Table 2Summary of outcomesCharacteristicN = 4,2781 Mortality at 30 postoperative days20 (0.5%) Mortality at 90 postoperative days88 (2.0%) Length of Stay (days)4.0 (3.0, 7.0)2 Any ICU stay93 (2.2%)Pationsostoperative Complic Pulmonary324 (7.6%) Stroke10 (0.2%) Cardiac Complication63 (1.5%) Thrombosis (including Pulmonary Embolism)53 (1.2%) Pulmonary Embolism27 (0.6%) Atrial Fibrillation/Flutter93 (2.2%) Sepsis39 (0.9%)1 n (%)2Median (IQR) Summary of outcomes 1 n (%) 2Median (IQR) Table 3 shows the results of univariable logistic regression analysis for the 30-day mortality endpoint. MINShsTnI was associated with increased odds of 30-day mortality (OR 3.3; 95% CI 1.07, 8.57, p = 0.039), but association with MINSTnI was not detected (OR 3.67; 95% CI 0.85, 11.0, p = 0.076). Several cardiovascular risk factors including congestive heart failure, pulmonary circulation disorders, chronic pulmonary disease, and hypertension with complications (as defined by the Elixhauser comorbidities) were found to be associated with increased mortality. Cardiotoxic cancer therapy was given within three months of surgery to 17% of the cohort (n = 743). This subset had increased odds of 30-day mortality (OR 3.20, 95% CI 1.25, 7.75, p = 0.017) and a higher MINSeither prevalence compared to non-exposed patients (12% vs. 9%; p = 0.015). Multivariable modeling was not performed due to the low event rate observed for 30-day mortality. Table 3Univariable logistic regression analysis for 30-day mortalityCharacteristic N 30-day mortality, NOR195% CI1p-value2Emergency status No4,23819—— Yes4015.690.31, 28.50.2Age (per year)4,278201.031.00, 1.07 0.047 Sex Male2,03214—— Female2,24660.390.14, 0.96 0.041 Duration of anesthesia (per hour)4,278201.080.92, 1.240.3Estimated blood loss (per 100 ml)4,278201.020.94, 1.050.5Cardiotoxic cancer therapy in prior 3 months No3,53512—— Yes74383.201.25, 7.75 0.017 Congestive Heart Failure No4,14815—— Yes130511.03.54, 28.9 < 0.001 Valvular Disease No4,09718—— Yes18122.530.40, 8.860.3Pulmonary Hypertension No4,22418—— Yes5428.991.41, 32.2 0.025 Chronic Pulmonary Disease No3,65812—— Yes62083.971.55, 9.64 0.005 Peripheral Vascular Disease No4,13618—— Yes14223.270.52, 11.50.2Hypertension No2,64013—— Yes1,63870.870.33, 2.120.8Hypertension with Complication No3,99714—— Yes28166.212.18, 15.6 0.001 Diabetes No3,86819—— Yes41010.500.03, 2.390.4Diabetes with Complication No4,12418—— Yes15423.000.47, 10.50.2Chronic Renal Disease No4,03417—— Yes24432.940.68, 8.830.13Atrial Fibrillation No4,01618—— Yes26221.710.27, 5.960.5Existence of Coronary Stent No4,11319—— Yes16511.310.07, 6.390.8Post-Operative Troponin Categorization MINSTnI4270203.670.85, 11.00.076 MINShsTnI4270203.301.07, 8.57 0.039 MINSeither4278203.211.04, 8.35 0.043 1OR Odds Ratio, CI Confidence Interval2Bolded p-values indicate p < 0.05 Univariable logistic regression analysis for 30-day mortality 1OR Odds Ratio, CI Confidence Interval 2Bolded p-values indicate p < 0.05 Secondary outcome analysis Observed to expected length of stay The median O: E LOS was higher in patients with MINShsTnI (median 1.10, IQR (0.75, 1.66)) than without (median 0.89, IQR (0.62, 1.28)). (Supplementary Table S4) MINShsTnI was associated with a statistically significantly higher observed to expected length of stay, an average increase of 25% (95% CI 16, 32%; p < 0.001) using univariable linear regression of the log-transformed ratio (Supplementary Table S5). Ninety-day mortality Analysis of 90-day mortality as a secondary outcome was performed using a Cox proportional hazard model. A significant association with MINShsTnI was not detected (HR 1.46, 95% CI 0.78, 2.71, p = 0.2) as shown in Table 4. Recent exposure to cardiotoxic therapeutics, emergent surgery, neurosurgery, and congestive heart failure risk factors were associated with significantly increased hazard ratios. Table 4Multivariable Cox proportional hazards model for 90-day mortalityCharacteristicsHR195% CI1p-value2Recent exposure to cardiotoxic therapeutics2.021.25, 3.25 0.004 Emergent Surgery6.712.85, 15.8 < 0.001 Peripheral Vascular Disease1.480.65, 3.390.4Chronic Pulmonary Disease1.230.70, 2.160.5Neurosurgery4.322.68,6.97 < 0.001 Atrial Fibrillation1.020.48, 2.18> 0.9Coronary Stent0.770.30, 1.990.6Congestive Heart Failure4.692.35, 9.37 < 0.001 Age1.021.00, 1.04 0.017 MINShsTnI1.460.78, 2.710.21HR Hazard Ratio, CI Confidence Interval2Bolded p-values indicate p < 0.05 Multivariable Cox proportional hazards model for 90-day mortality 1HR Hazard Ratio, CI Confidence Interval 2Bolded p-values indicate p < 0.05 Exploratory analysis for HsTnI risk categorization The relationship between postoperative hsTnI risk category [low, intermediate, high thresholds, based on sex] and 90-day mortality are summarized in Table 5. Two patients were excluded from this analysis who died within 3 days after surgery based on T0 definition. Using the threshold for low cardiovascular risk in the general population (< 6 and < 4 ng/L for males and females, respectively) as the reference group, patients with intermediate risk hsTnI concentration that was below the 99th percentile had an estimated hazard ratio of 1.84 (95% CI 1.17, 2.91; p = 0.009), while patients above the 99th percentile had a hazard ratio of 2.05 (95% CI 1.07, 3.91; p = 0.030). Using low risk as the reference group, patients with intermediate risk hsTnI concentration had a hazard ratio of 1.65 (95% CI 0.98, 2.78; p = 0.061), while patients with high risk hsTnI concentration had a hazard ratio of 2.17 (95% CI 1.31, 3.58; p = 0.003). Table 5Univariable Cox proportional hazards models for 90-day mortality by HsTnI risk categoryCharacteristic N 90–day mortality, N2HR195% CI1p-valuePeak POD hsTnI risk category Low (< 6 ng/L male, < 4 ng/L female)2,60539—— Intermediate & below 99th percentile1,270351.841.17, 2.91 0.009 ≥ 99th Percentile (> 35 ng/L male, >17 ng/L female)393122.051.07, 3.91 0.030 Peak POD hsTnI risk category Low (< 6 ng/L male, < 4 ng/L female)2,60539—— Intermediate (6–12 ng/L male, 4–10 ng/L female)889221.650.98, 2.780.061 High (> 12 ng/L male, >10 ng/L female)774252.171.31, 3.58 0.003 To defined as 3 days post-surgery (landmark analysis for the latest post-operative draw), maximum follow-up = 90 days after surgery procedure1 HR Hazard Ratio, CI Confidence Interval2 Two patients excluded as they both died 3 days after surgery (which is the start of the follow up period Univariable Cox proportional hazards models for 90-day mortality by HsTnI risk category To defined as 3 days post-surgery (landmark analysis for the latest post-operative draw), maximum follow-up = 90 days after surgery procedure 1 HR Hazard Ratio, CI Confidence Interval 2 Two patients excluded as they both died 3 days after surgery (which is the start of the follow up period Clinician-ordered troponin-I Clinicians ordered TnI testing independent of the study for 284 patients (6.6%). Of the patients with MINSeither by our testing, 18% (73 of 405 patients) had a clinician ordered TnI assay during their admission. 10% (40 of 405) of patients with MINSeither were detected by clinician-ordered troponin testing (> = 0.06 ng/mL or 60 ng/L). Discussion In univariable analysis, this study demonstrated an association between MINS measured using high sensitivity troponin-I and 30-day mortality in a cancer surgery population. An association between MINS and 90-day mortality, however, was not detected in multivariable Cox analysis. Our analyses also found that recent cardiotoxic cancer therapies were associated with increased prevalence of MINS as well as univariable association with 30-day mortality. The importance of detecting MINS was also emphasized by an association between MINS and increased observed to expected length of stay ratio, which indicates that MINS has an impact on length of stay. 30-day mortality association In our population, MINS detected by high-sensitivity troponin-I (MINShsTnI) was associated with 3.3-fold greater odds of 30-day mortality in univariable analysis (95% CI 1.07, 8.57). This is consistent with previous studies of myocardial injury using other high sensitivity troponin assays, but contrasts with a study of thoracic surgery patients that showed no association with 30-day mortality [4, 20]. The prevalence of MINShsTnI was 9.3% in our study of postoperative cancer patients. Landmark studies of postoperative patients in the general population reported a MINS prevalence of 8% (using TnT) [5] and 17.9% (using hsTnT) [4]. van Waes et al. (2013) found a MINS prevalence of 19.4% using a contemporary TnI assay [21]. Similar rates were found in patients after thoracic surgery (27%), carotid endarterectomy (15.3%) and endovascular aneurysm repair (16.2%) [20, 22, 23]. The lower prevalence of MINS in our study population may relate to fewer baseline comorbidities compared to the 2017 VISION study using hsTnT. Notable differences in comorbidities include diabetes mellitus (9.6% vs. VISION 20.6%), hypertension (38% vs. VISION 49.8%), and peripheral vascular disease (3.3% vs. VISION 9.3%) [4]. Conversely, rates of active cancer (100% vs. VISION 24%) and COPD (14% vs. VISION 7.5%) were higher within our study cohort [4]. Another possible explanation for the lower mortality rate is the younger age limit for inclusion. Of our population, 17% were younger than 45 years old and would have been excluded from the VISION study based on age. However, the age distribution is similar despite the younger patients. In our study, 60% were less than 65 years old and 14% greater than or equal to 75 years old versus the 2017 VISION study which had 58.5% and 15.5% respectively [4]. Additional factors that may contribute to outcome differences among studies include heterogeneity in troponin assays and thresholds used. There are few studies in the cancer surgery setting examining the link between troponin results and postoperative mortality. One study in patients undergoing GI cancer surgery found that an elevated preoperative hsTnI was one of the strongest predictors of mortality and/or severe cardiac complication endpoints (odds ratio (OR) 5.924 (95%CI: 2.869–12.233), p < 0.001), but this study did not examine postoperative hsTnI [24]. Cardiotoxic cancer therapy association with 30-day mortality This study found an association of 30-day mortality in patients who had cardiotoxic cancer therapies within 3 months prior to surgery. This is a previously unreported association. In addition, these patients also had a higher prevalence of MINShsTnI compared to non-exposed patients. This association is biologically plausible because many agents are linked to subclinical myocardial injury. Anthracyclines have been associated with early diastolic dysfunction and microvascular dysfunction [25]. Small molecule kinase inhibitors can impair endothelial function and trigger apoptosis in cardiac myocytes [26]. HER2 inhibitors predispose cardiac myocytes to mitochondrial stress and cellular injury [27]. Several studies have suggested that troponin elevation during anthracycline and/or trastuzumab therapy may indicate a higher risk for future development of cardiotoxicity [28, 29]. Currently, patients receiving such therapies may only have troponin levels checked when symptomatic even though routine troponin surveillance in asymptomatic patients has been recommended for patients undergoing treatment with potentially cardiotoxic anticancer agents (e.g., anthracyclines, trastuzumab) in recent guidelines [12]. However, cardiotoxic cancer therapy could be a confounder since these patients are more likely to have advanced cancer which is associated with increased mortality. Relationship between MINS and increased length of stay The ratio of observed to expected length of stay was significantly higher for patients with MINS with an average increase of 25% (95% CI 16, 32%) in the O: E LOS for MINShsTnI. A longer than expected hospital stay can be due to patient factors such as comorbidity burden or hospital factors such as discharge bottlenecks. The association between MINS and increased length of stay was found despite the fact that clinicians were blinded to the troponin results. This association suggests that MINS may be a surrogate marker of a complicated perioperative course. MINS can coincide with postoperative hypertension and renal dysfunction. MINS patients may have had subclinical signs such as mild hypotension or dyspnea that led to additional testing or observation which delayed discharge. Comparison of using hsTnI and TnI in MINS An important consideration for any MINS study is selection of thresholds for troponin elevation because there is currently no universally accepted MINS threshold for hsTnI assays. Different hsTnI reagent vendors have different capture and detection epitopes, and different reference ranges within healthy populations, and therefore direct comparisons between vendor assays can be complex and should be avoided. Furthermore, hsTnI reference ranges are different for females versus males [30]. This study used vendor-defined sex-specific thresholds, which have been demonstrated to provide additional insight in suspected MI, compared to an overall threshold which can result in the underdiagnosis of MI in the acute setting, especially in females [31]. However, there is literature to support using an overall threshold for MINS [32, 33]. Primary analysis was performed using a non-sex specific MINS threshold of 28 ng/mL and the results were largely unchanged (Supplementary Table S6). By comparison, the VISION study used an overall threshold for hsTnT. Although, sex specific thresholds have been defined, hsTnT assays do not consistently detect troponin T in at least 50% of females in a healthy reference population [34, 35]. In this study, the hsTnI assay demonstrated superior performance at detecting MINS than the TnI assay (9% of patients vs. 4.6% of patients). MINS as defined by hsTnI was associated with increased odds of 30-day mortality whereas TnI was not. When comparing TnI and hsTnI, it is important to note that with each successive generation of troponin assay technology, sensitivity and precision have improved. Contemporary TnI assays cannot reliably detect troponin in the majority of healthy patients. Because the contemporary assay’s limit of detection exceeded the 99th percentile, MINSTnI was diagnosed if any detectable troponin level was measured. These differences must be appreciated when comparing troponin results and MINS data generated using these assays. HsTnI risk categorizations in the exploratory analysis showed a statistically significant association with 90-day mortality in a univariable model (Table 5), although MINS defined as the hsTnI 99th percentile was not statistically significant in a multivariable model of 90-day mortality (Table 4). These troponin thresholds (low/intermediate/high) linked to cardiovascular risk were based on research in an asymptomatic general population [16]. Both intermediate risk hsTnI and high risk hsTnI thresholds had significantly elevated hazard ratios (1.65 and 2.17 respectively) compared to the low risk hsTnI threshold used as a reference group (Table 5). The high risk threshold (12 ng/L male and 10 ng/L female) is much lower than the 99th percentile threshold (35 ng/L male and 17 ng/L female), emphasizing that relying on the 99th percentile threshold may lead to missed clinically significant cardiac injury. The cost-effectiveness of mass postoperative hsTnI surveillance was not an outcome studied. Based on our findings we believe that focusing on troponin testing in patients with risk factors such as cardiotoxic chemotherapy would have a favorable cost benefit ratio. Prior work using data from Canadian patients in the VISION study found that the incremental cost to avoid missing a MINS event in patient subgroups at higher risk for MINS was $1,309 in 2015 Canadian dollars, versus $1,602 in all patients studied [36]. Strengths and limitations of study One of the major strengths of this study is the prospective study design in a large cancer patient population. Cancer patients have unique considerations that may affect myocardial injury during surgery. There is an increased benefit of the prognostic value of MINS in cancer patients since these patients need to address challenging goals-of-care decision-making. Although our study is based on a single cancer-specific institution with a lower comorbidity burden than the VISION studies, this patient population supports that these results reflect cancer-specific risks for MINS. The applicability of these results for community hospitals or less complex oncologic surgeries may be more limited. Another major strength is that this study presents a head-to-head comparison of hsTnI versus cTnI to detect MINS. The high sensitivity troponin assay used in this study meets IFCC criteria that the test detect troponin in over half of the reference population [34]. The low event rate observed for the primary outcome of 30-day mortality is a significant limitation of this study. Despite attempting to include more complex, inpatient surgeries, the mortality rate of the study population was lower than expected by the initial power analysis. The prevalence of MINS may be underestimated because waste blood was not collected for all patients at each desired timepoint. However, all patients had at least one measurement and half of the population had at least three postoperative samples. The vital status of 17 patients could not be determined within 30 post-operative days and were assumed to be alive. The primary outcome of this study was all-cause 30-day mortality rather than cardiac death. All-cause mortality remains the standard for establishing the effectiveness of perioperative biomarkers. Although similar outcomes have been used from other MINS studies [2, 4, 5], one might argue that cardiac death is more relevant when investigating an association with MINS. All-cause mortality is a more comprehensive and relevant outcome to cancer patients. The clinical significance of cardiac complications cannot be easily separated from the underlying oncologic trajectory. Using a cardiac-specific mortality endpoint would miss situations where MINS led to palliation, which would attribute a death to cancer instead of heart damage. These downstream consequences would not be fully captured by a cardiac-specific mortality endpoint. Inclusion of non-cardiac mortality can dilute the association between MINS and mortality, so our results underestimate the strength of that association. This patient population were prospectively identified as inpatient surgeries, but 22% of the patients had a LOS less than or equal to two days. This may represent an underestimation of surgical complexity with surgical scheduling practices possibly favoring overestimating length of stay to ensure an inpatient bed was reserved. An increase in observed-to-expected LOS ratio reflects a longer than predicted hospital course. Clinically, this often signifies a higher burden of postoperative complications and delayed recovery, which are both associated with increased morbidity. Furthermore, prolonged LOS directly translates to increased healthcare utilization and costs from extended stay and downstream complications. Therefore, observed-to-expected LOS ratio serves as a practical surrogate for both morbidity and healthcare expenditure. Conclusions Our findings suggest there may be a role for postoperative hsTnI surveillance for MINS in the cancer surgery population. Comorbidities such as age, congestive heart failure, and recent cardiotoxic cancer therapy may identify a smaller subset of patients who benefit from MINS surveillance. Patients who are identified with MINS can be treated with an antithrombotic agent such as aspirin [37]. The expectation is that surveillance and treatment of MINS will reduce 30-day mortality and length of stay. The finding that cancer patients with a cardiotoxic cancer therapy within 3 months of surgery had three-fold higher odds of 30-day mortality is a new contribution to the MINS literature. The mechanism of interaction for MINS and preceding cardiotoxic cancer therapies has not been examined and more studies in this area will be needed. Supplementary Information Supplementary Material 1 Acknowledgements Data from the Vizient Clinical Data Base used by permission of Vizient, Inc. All rights reserved. Authors’ contributions T.L. Contributions: Conceptualization, Methodology, Validation, Formal Analysis, Investigation, Resources, Data Curation, Writing-original draft, Writing-review and editing, Visualization, Supervision, Project Administration, Funding acquisitionP.J.M. Contributions: Methodology, Validation, Formal Analysis, Investigation, Resources, Data Curation, Writing-original draft, Writing-review and editing, Visualization, Supervision, Project Administration; D.A. Contributions: Conceptualization, Methodology, Writing-review and editing; K.S.T. Contributions: Conceptualization, Validation, Formal Analysis, Resources, Data Curation, Writing-original draft, Writing-review and editing, Visualization, Funding acquisition; J.L. Contributions: Methodology, Validation, Formal Analysis, Resources, Data Curation, Writing-original draft, Writing-review and editing, Visualization; G.W.F. Contributions: Conceptualization, Methodology, Writing-review and editing, Supervision, Funding acquisitionL.J. Contributions: Analysis, Writing-original draft, Writing-review and editing; D.D. Contributions: Writing-review and editing; L.V.R. Contributions: Conceptualization, Methodology, Writing-review and editing, Supervision, Funding acquisition; C.L.C. Contributions: Conceptualization, Methodology, Writing-review and editing; G. M. Contributions: Methodology, Resources, Writing-review and editing, Funding acquisition; A.G.B. Contributions: Conceptualization, Methodology, Validation, Formal Analysis, Investigation, Resources, Data Curation, Writing-original draft, Writing-review and editing, Visualization, Supervision, Project Administration, Funding acquisition. Funding The research was supported by the National Cancer Institute of the National Institutes of Health under Award Number P30CA008748. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. The TnI and hsTnI kits were supplied by Abbott Diagnostics. Data availability The dataset for this study is available on reasonable request from the corresponding author. Declarations Ethics approval and consent to participate This study was conducted in accordance with the Declaration of Helsinki. The study was approved by the Institutional Review Board (IRB) of Memorial Sloan Kettering Cancer Center (approval IRB# 17–643). The need for informed consent from individual patients was waived by the IRB. Clinical trial number: not applicable. Consent for publication Not applicable. Competing interests P.J.M.‘s spouse holds stock in Johnson & Johnson. G.M., D.D., and L.J. are full time employees of Abbott Laboratories. Footnotes Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary Material 1 Data Availability Statement The dataset for this study is available on reasonable request from the corresponding author.