Mortality risk prediction of high-sensitivity C-reactive protein in suspected acute coronary syndrome: A cohort study.

BACKGROUND: There is limited evidence on the use of high-sensitivity C-reactive protein (hsCRP) as a biomarker for selecting patients for advanced cardiovascular (CV) therapies in the modern era. The prognostic value of mildly elevated hsCRP beyond troponin in a large real-world cohort of unselected patients presenting with suspected acute coronary syndrome (ACS) is unknown. We evaluated whether a mildly elevated hsCRP (up to 15 mg/L) was associated with mortality risk, beyond troponin level, in patients with suspected ACS. METHODS AND
FINDINGS: We conducted a retrospective cohort study based on the National Institute for Health Research Health Informatics Collaborative data of 257,948 patients with suspected ACS who had a troponin measured at 5 cardiac centres in the United Kingdom between 2010 and 2017. Patients were divided into 4 hsCRP groups (<2, 2 to 4.9, 5 to 9.9, and 10 to 15 mg/L). The main outcome measure was mortality within 3 years of index presentation. The association between hsCRP levels and all-cause mortality was assessed using multivariable Cox regression analysis adjusted for age, sex, haemoglobin, white cell count (WCC), platelet count, creatinine, and troponin. Following the exclusion criteria, there were 102,337 patients included in the analysis (hsCRP <2 mg/L (n = 38,390), 2 to 4.9 mg/L (n = 27,397), 5 to 9.9 mg/L (n = 26,957), and 10 to 15 mg/L (n = 9,593)). On multivariable Cox regression analysis, there was a positive and graded relationship between hsCRP level and mortality at baseline, which remained at 3 years (hazard ratio (HR) (95% CI) of 1.32 (1.18 to 1.48) for those with hsCRP 2.0 to 4.9 mg/L and 1.40 (1.26 to 1.57) and 2.00 (1.75 to 2.28) for those with hsCRP 5 to 9.9 mg/L and 10 to 15 mg/L, respectively. This relationship was independent of troponin in all suspected ACS patients and was further verified in those who were confirmed to have an ACS diagnosis by clinical coding. The main limitation of our study is that we did not have data on underlying cause of death; however, the exclusion of those with abnormal WCC or hsCRP levels >15 mg/L makes it unlikely that sepsis was a major contributor.
CONCLUSIONS: These multicentre, real-world data from a large cohort of patients with suspected ACS suggest that mildly elevated hsCRP (up to 15 mg/L) may be a clinically meaningful prognostic marker beyond troponin and point to its potential utility in selecting patients for novel treatments targeting inflammation. TRIAL REGISTRATION: ClinicalTrials.gov - NCT03507309.

Introduction

Inflammation has been extensively studied and postulated as mechanistic in atherothrombosis [1]. Much debate has been on whether the immune system plays a protective or pathogenic role in atherogenesis and subsequent cardiovascular (CV) events. In addition to exploiting elements of the immune system as targets for novel therapies, some have been investigated for their role as potential biomarkers, aiming to permit better cardiovascular disease (CVD) risk stratification [2,3]. C-reactive protein (CRP) is the most widely evaluated of these biomarkers, given its relatively long circulating half-life and commonplace use in clinical practice.

In early observational studies of general and acute coronary syndrome (ACS) populations, high levels of CRP were found to represent a poorer prognosis, perhaps underlying a previously unrecognised inflammatory process relating to atherosclerosis [2,4-6]. Further studies have added significant weight to these observations, identifying that baseline serum CRP levels are an independent predictor of incident CV events and mortality in healthy individuals [7,8] as well as in patients with established CVD [9-12]. Although CRP has been shown to be a useful correlative biomarker in primary and secondary prevention, it is likely to be an end-protein readout of a mechanistic pathway involving interleukin (IL)-6, rather than a therapeutic target itself [13,14].

Considering high-sensitivity C-reactive protein (hsCRP) levels in a stable populations as a biomarker for selecting patients for advanced CV therapies in the modern era was pioneered by the Canakinumab Anti-inflammatory Thrombosis Outcome Study (CANTOS), a randomised, double-blind, placebo-controlled trial of canakinumab (a human monoclonal antibody targeted against IL-1β, upstream of IL-6), in preventing recurrent CV events. Patients in the canakinumab arm had a significant 15% reduction in CVD events over median 3.7 years of follow up, coexistent with reduced hsCRP levels and independent of lipid lowering [15]. Moreover, those that failed to achieve on-treatment hsCRP <2 mg/L with canakinumab saw no improvement in major adverse cardiac events (MACE) [16].

Results of the Cardiovascular Inflammation Reduction Trial (CIRT) had shed some doubt onto the role of modulating inflammation in atherosclerosis using rheumatological agents, after it terminated early and reported that low-dose methotrexate had no effect on CVD versus placebo [17]. However, this was not restricted to those with elevated hsCRP at baseline, suggesting that there may still be a significant treatment gap of focused inflammatory targeting in atherosclerosis. Somewhat surprisingly, no effects of methotrexate were seen on IL-6, IL-1β, or hsCRP. More recent supportive evidence of targeting inflammation in CVD has been published, with the Colchicine Cardiovascular Outcomes Trial (COLCOT) [18] and the Low Dose Colchicine (LoDoCo2) for secondary prevention of CVD trial [19] reporting significant reductions in CV end points using colchicine, an immunomodulatory agent that works on several pathways including targeting IL-1β, in patients within 30 days of ACS or with chronic coronary disease. However, laboratory measures of inflammation (including CRP) at baseline from these studies are not available. If the baseline absolute risk of patients with ACS and elevated CRP values is high, then targeting these individuals with anti-inflammatory therapies will have larger clinical impact with lower values for the number needed to treat.

However, in patients presenting with suspected ACS, levels of hsCRP may either reflect background atherosclerotic disease activity or indeed also reflect the sequelae of the acute event. Nevertheless, hsCRP levels at presentation may prove useful in prognostication beyond routinely studied parameters, including troponin.

In this study, we aim to evaluate the prognostic value of mildly elevated hsCRP (up to 15 mg/L) beyond troponin in a large real-world cohort of unselected patients presenting with suspected ACS. Our hypothesis was that a mildly elevated hsCRP would be associated with mortality risk, beyond troponin, in patients with suspected ACS.

Methods

We conducted a retrospective cohort study based on the National Institute for Health Research Health Informatics Collaborative data [20-22] of 257,948 patients with suspected ACS who had a troponin measured at 5 cardiac centres in the UK between 2010 and 2017 (Imperial College Healthcare, University College Hospital, Oxford University Hospital, Kings College Hospital and Guys and St Thomas’‘ Hospital). Ethical approval for the dataset was obtained. The data acquisition and statistical analysis plan are available as S1 Data Aquisition and S1 Analysis Plan, respectively. There was no deviation in the statistical analysis plan during conduct of the study or following peer review. The study was registered at ClinicalTrials.gov, NCT03507309.

We developed a data model to capture the longitudinal record for patients with a suspected ACS, characterised by the request of a troponin test. For patients with multiple episodes of care for which troponin was tested, the first episode of care was used. We intentionally focused our population to those with normal white cell count (WCC) in an attempt to exclude overtly septic patients or those with haematological disorders that may have a major effect on levels of hsCRP. Furthermore, we excluded all patients with hsCRP >15 mg/L in a separate effort to exclude those with clinically significant inflammation or infection. All eligible patients were followed up, using routinely collected data on the National Health Service (NHS) Spine Application, until death or censoring in April 2017.

All analyses on hsCRP and other haematological and biochemical blood tests, apart from troponin, were performed using the first result measured during the hospital care episode.

The 99th percentile of the upper limit of normal (ULN) and the limit of detection of all hsCRP assays used across all cardiac centres were determined.

All analyses on troponin were performed using the peak troponin level. For patients who had a single troponin measurement, the peak troponin was based on this measurement. In the remainder of the patients who had more than one troponin test in the same hospital episode of care, the peak troponin value was defined as the highest of all measurements. We standardised the different troponin assays between the academic centres by scaling the results using the ratio of the observed troponin value divided by the ULN for each troponin assay. Each centre measured troponin I or troponin T using either contemporary or high-sensitivity assays (S1 Table). These tests yielded results in different measurable ranges, with unique cutoff points for the ULN. There is current lack of standardisation of the various cardiac troponin assays as a physiological threshold for myocardial injury [23,24]. The ULN provided by the manufacturer for each troponin assay provides an analytical cutoff between normal and increased cardiac troponin levels. A positive troponin was defined as a result above the ULN for each troponin assay.

Patients who were admitted to the hospital had the International Classification of Diseases (ICD) discharge codes. Patients were classified as ACS based on the assigned ICD-10 codes (S2 Table) [25].

The population was divided into 4 categories, those with a normal hsCRP (<2 mg/L) [15] and 3 groups of incrementally increasing hsCRP (2 to 4.9 mg/L, 5 to 9.9 mg/L, and 10 to 15 mg/L). The groupings were selected to reflect low grade inflammation (overt infections were also excluded using the WCC data). The median baseline hsCRP in CANTOS was 4.2 mg/L; therefore, a smaller group of 2 to 4.9 mg/L was chosen to examine this cohort more closely [15]. The 4 hsCRP groups were further subdivided according to troponin positivity and subgroups including only patients diagnosed with an ACS during their index hospital admission.

This study was approved by the London South East Research Ethics Committee (REC reference: 16/HRA/3327). This study is reported as per the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guideline (S1 Checklist).

Statistical analysis

Descriptive statistics are displayed as median (interquartile range (IQR)) for continuous variables and number (%) for categorical variables. Comparisons between hsCRP groups with regard to baseline characteristics were explored using Kruskal–Wallis 1-way analysis of variance test and χ2 test for trend for continuous and categorical variables, respectively. The correlation between hsCRP and troponin concentrations was assessed using the Spearman correlation coefficient. The Kaplan–Meier method was used to calculate and display cumulative mortality, with hsCRP groups compared using the log-rank statistic. Subgroup analyses were performed for troponin positivity and ACS diagnosis.

Multivariable Cox regression analysis was applied to investigate whether the baseline hsCRP groups independently predict 3-year mortality after adjusting for other demographic and clinical variables. The proportional hazard assumption was violated with a significant relationship between the Schoenfield residuals of all covariates and time. Cox regression analysis with time-varying coefficients was therefore used with follow-up time divided into 6 intervals: <1 month, 1 to 3 months, 3 to 6 months, 6 to 12 months, 12 to 24 months, and 24 to 36 months. The proportional hazard assumption was met in each of these time intervals. The coefficients calculated during each of these time intervals were based on the variables measured during the index hospital admission. Furthermore, using Martingale residuals, nonlinearity was detected in the relationship between the log hazard and all continuous covariates. To model these nonlinear relationships, we used restricted cubic splines in the Cox proportional hazards model.

To assess the predictive role of hsCRP on short-term (30 days) and long-term (3 years) mortality beyond conventional risk factors and troponin, we established 3 statistical models. Model 1 was adjusted for age, sex, haemoglobin, and creatinine; model 2 was additionally adjusted for troponin positivity; model and 3 was further adjusted for hsCRP (hsCRP <2 mg/L, 2 to 4.9 mg/L, 5 to 9.9 mg/L, and 10 to 15 mg/L). The statistical models were based on the predicted probabilities of a logistic regression model. Ethnicity was not included in the risk model due to approximately 10% of patients having missing data. Furthermore, for those who had ethnicity recorded, approximately 15% had “other” recorded as an ethnic category, therefore making it less easy to interpret as part of a clinical risk model.

Discrimination of the different models was assessed by comparing the areas under the receiver operating characteristic curves (AUROCs) using the DeLong nonparametric approach. We also computed the continuous net reclassification index (NRI) and performed an integrated discrimination improvement (IDI) analysis with the censored survival data. Comparing the 3 models sequentially, the NRI reflects the net increment in prediction accuracy, while the IDI reflects the change in calculated risk for each individual. The IDI reflects the change in discrimination slope (i.e., difference between the mean estimated risk for patients who died and those who survived) of the model with the new predictor compared to the model with only the established predictors. IDI (range −100% to 100%) represents overall risk discrimination improvement.

In order to evaluate the performance of negative hsCRP and troponin values in predicting mortality, the negative predictive values of a negative hsCRP, negative troponin, or both a negative troponin and hsCRP were plotted against the range of hypothetical mortality rates at 30 days and 3 years. The negative predictive value, which varies with the prevalence of disease (mortality), is defined as the probability of not dying when the test result is negative.

Statistical analyses were performed using R 3.5.0 (R Core Team, Vienna, Austria) or MedCalc version 15.8 (MedCalc Software, Mariakerke, Belgium). The survIDINRI package on R was used to implement the IDI and NRI for comparing risk prediction models.

Results

Baseline population characteristics

A total of 257,948 patients presented with suspected ACS to 5 acute cardiac centres and were included in the NIHR Health Informatics Collaborative dataset. Moreover, 71,262 patients did not undergo hsCRP testing during their hospital care episode and were therefore excluded from further analysis. A further 67,206 patients were excluded based on hsCRP level >15 mg/L. The remaining 129,480 patients were divided into 4 groups by hsCRP levels, reflecting low grade inflammation. After further exclusion of patients with abnormal WCCs (<4 or >11 × 109/L) or those who did not have a WCC blood test, 102,337 patients were included in the final analysis. The median follow-up was 3.3 years (IQR 1.4 to 5.1). The study cohort was divided into 4 hsCRP categories (hsCRP <2 mg/L (n = 38,390), 2 to 4.9 mg/L (n = 27,397), 5 to 9.9 mg/L (n = 26,957), and 10 to 15 mg/L (n = 9,593)). Overall, 53.3% (n = 54,534) of patients in the study cohort were admitted to the hospital. Among these patients, 10.8% (n = 5,910) were coded with an ACS during their inpatient stay. Patient flow through the study is shown in Fig 1. S3 Table displays the limits of detection and references ranges for the 99th percentile of the ULN for the hsCRP assays used in the cardiac centres included in the study. Of note, all assays had a lower limit of detection ≤0.2mg/L with assays validated internally by all institutions.

Fig 1

Flow of patients through the study.

*Suspected ACS characterised by the request of a troponin. ACS, acute coronary syndrome; hsCRP, high-sensitivity C-reactive protein; ICD, International Classification of Diseases; WCC, white cell count.

Table 1 displays the baseline clinical characteristics for the patients included in the analysis. Patients with higher hsCRP values were older and more likely to be female. Although there were significant differences in levels of troponin between hsCRP groups, there was no obvious trend. We tested this relationship further to establish if the variables were interdependent, finding a very weak correlation with Spearman coefficient R = 0.15 (p < 0.001) (S1 Fig). Although we tested hsCRP in relation to “peak” troponin available in predicting mortality, we established that for patients with an elevated troponin result, the median number of days between the first admission hsCRP measured and peak troponin level was 3 (IQR 0 to 5) days, suggesting that the admission hsCRP represents presentation values rather than the sequelae of the index event.

Table 1

Baseline clinical characteristics.

	hsCRP <2 mg/L (n = 38,390)		hsCRP 2 to 4.9 mg/L (n = 27,397)		hsCRP 5 to 9.9 mg/L (n = 26,957)		hsCRP 10 to 15 mg/L (n = 9,593)		p-Value
	n (median (IQR) or number (% of total patients))	Total patients (% of all patients)	n (median (IQR) or number (% of total patients))	Total patients (% of all patients)	n (median (IQR) or number (% of total patients))	Total patients (% of all patients)	n (median (IQR) or number (% of total patients))	Total patients (% of all patients)
General demographics
Age (years)	59 (45 to 75)	38,374 (99.96)	65 (51 to 79)	27,381 (99.94)	64 (50 to 79)	26,952 (99.98)	70 (54 to 82)	9,589 (99.96)	<0.001
Male	21,566 (56.20)	38,374 (99.96)	14,377 (52.51)	27,378 (99.93)	14,292 (53.03)	26,951 (99.98)	4,774 (49.79)	9,589 (99.96)	<0.001
Ethnicity
White	22,824 (71.04)	32,130 (83.69)	17,157 (73.69)	23,284 (84.99)	16,698 (70.54)	23,671 (87.81)	6,095 (74.37)	8,195 (85.43)	<0.001
South Asian	1,445 (4.50)		993 (4.26)		879 (3.71)		304 (3.71)
Black	3,018 (9.39)		2,197 (9.44)		3,367 (14.22)		960 (11.71)
Other	4,843 (15.07)		2,937 (12.61)		2,727 (11.52)		836 (10.20)
Haematology/biochemistry
Haemoglobin (g/dL)	13.8 (12.7 to 14.9)	38,389 (100)	13.6 (12.4 to 14.7)	27,395 (99.99)	13.4 (12.1 to 14.5)	26,954 (99.99)	13.0 (11.7 to 14.2)	9,592 (99.99)	<0.001
WCC (×109/L)	7.1 (5.9 to 8.5)	38,390 (100)	7.5 (6.2 to 8.9)	27,397 (100)	7.6 (6.3 to 9.0)	26,957 (100)	7.9 (6.5 to 9.3)	9,593 (100)	<0.001
Platelet count (×109/L)	225 (190 to 264)	38,373 (99.96)	230 (192 to 272)	27,391 (99.98)	226 (185 to 270)	26,941 (99.94)	230 (186 to 281)	9,589 (99.96)	<0.001
Creatinine (μmol/L)	74 (64 to 88)	38,279 (99.71)	76 (65 to 92)	27,338 (99.78)	76 (64 to 94)	26,882 (99.72)	78 (65 to 100)	9,584 (99.91)	<0.001
Positive troponin	6,099 (15.9)	38,390 (100)	5,937 (21.7)	27,397 (100)	7,402 (27.5)	26,957 (100)	3,126 (32.6)	9,593 (100)	<0.001
Troponin level (×ULN)	0.003 (0.003 to 0.4)	38,390 (100)	0.003 (0.003 to 0.7)	27,397 (100)	0.23 (0.003 to 1.25)	26,957 (100)	0.06 (0.003 to 1.7)	9,593 (100)	<0.001

p-Values were calculated using Kruskal–Wallis 1-way analysis of variance and χ2 test for trend for continuous and categorical variables, respectively.

hsCRP, high-sensitivity C-reactive protein; IQR, interquartile range; ULN, 99th percentile of the upper limit of normal; WCC, white cell count.

Higher hsCRP predicts significantly increased mortality risk

Patients in all 4 hsCRP categories had significantly incremental cumulative mortality over 3 years, in line with increasing hsCRP at baseline. Fig 2A displays cumulative mortality per hsCRP group, revealing increasing mortality with each consecutive group. Fig 2B further stratifies the groups according to dichotomised peak troponin level as positive or negative. This shows the greatest mortality for patients in the highest hsCRP group who also had a positive troponin assay (36.0% at 3 years). The additive effect of hsCRP to troponin on mortality is seen in all groups. There is also a graded mortality increase seen across the hsCRP troponin-negative groups, with lower mortality at 3 years than the troponin-positive groups. However, the troponin-negative highest hsCRP group (10 to 15 mg/L) had greater mortality at 3 years than the troponin-positive lowest hsCRP group (<2 mg/L) (p < 0.001). A similar relationship with hsCRP is seen in patients with different levels of troponin positivity (S2 Fig). We have Supporting information available on final ACS discharge diagnosis from clinical coding. Although these data have low granularity and are not available for all patients, we were able to use it to test if the relationship between hsCRP was still valid if we divided the population on discharge diagnosis of ACS, rather than suspected ACS on admission. Fig 2C demonstrates that this graded relationship between hsCRP and mortality was still present when restricting the cohort to just those clinically coded with confirmed ACS. A positive relationship between hsCRP level and mortality was also observed in patients admitted to the hospital without a clinically coded diagnosis of ACS (S3 Fig).

Fig 2

Unadjusted Kaplan–Meier mortality curves by (A) hsCRP level, (B) hsCRP level and troponin positivity, and (C) hsCRP level in a subgroup of patients with ACS. ACS, acute coronary syndrome; hsCRP, high-sensitivity C-reactive protein; Tn +, troponin positive, Tn −, troponin negative.

When undertaking Cox regression analysis with time-varying coefficients, hsCRP, albeit mildly raised, was an independent predictor of mortality over time, after adjusting for available clinically relevant covariates: age, sex, haemoglobin, WCC, platelet count, creatinine, and troponin positivity (Fig 3). At 30-day follow-up, the effect of even a very modestly raised hsCRP is evident, demonstrating a graded relationship with mortality (hazard ratio (HR) 1.27 (95% CI 1·10 to 1·48), HR 1.44 (95% CI 1.25 to 1.66), and HR 2.20 (95% CI 1.88 to 2.57) for hsCRP groups 2 to 4.9 mg/L, 5 to 9.9 mg/L, and 10 to 15 mg/L respectively, relative to a normal CRP level (<2 mg/L)). This effect persisted up to 3 years in the study, with HR (95% CI) of 1.32 (1.18 to 1.48) for those with hsCRP 2.0 to 4.9 mg/L, 1.40 (1.26 to 1.57), and 2.00 (1.75 to 2.28) for those with hsCRP 5 to 9.9 mg/L and 10 to 15 mg/L, respectively. The HRs for mortality risk at 3 years in the hsCRP groups stratified by troponin positivity are shown in S4 Table.

Fig 3

Multivariable Cox regression analysis with time-varying coefficients.

HRs are adjusted for age, sex, haemoglobin, WCC, platelet count, creatinine, and troponin level. The adjusted HRs were calculated during the following time periods: <1 month, 1 to 3 months, 3 to 6 months, 6 to 12 months, 12 to 24 months, and 24 to 36 months. The HRs are displayed at the end of each respective time period. The error bars depict 95% CIs. HR, hazard ratio; hsCRP, high-sensitivity C-reactive protein; WCC, white cell count.

hsCRP significantly improves mortality risk prediction

We next explored whether inclusion of hsCRP could better reclassify the population into at-risk mortality groups. The association with short-term and long-term mortality was assessed using 3 different risk models (model 1: age, sex, haemoglobin, and creatinine; model 2: model 1 plus troponin (positivity versus negativity); and model 3: model 2 plus hsCRP groups (<2, 2 to 4.9, 5 to 9.9, and 10 to 15 mg/L) (Table 2). For cumulative mortality at each time point, each successive model was better able to discriminate risk than its precursor (p < 0.001), such that inclusion of troponin and hsCRP gave the most robust risk discrimination. Of note, model 3, which includes both troponin positivity and hsCRP categories, achieves an AUROC >0.8 at 30 days and 3-year mortality, surpassing the use of troponin on its own. The addition of model 3 over model 2 resulted in an IDI of 0.3% and 0.9% at 30 days and 3 years, respectively (all p < 0·001), as well as the highest overall NRI of 18.4% and 13.9% at the same time points, respectively (all p < 0·001). Thus, the inclusion of hsCRP to the basic model with troponin level significantly improved all-cause mortality risk prediction.

Table 2

hsCRP risk model discrimination, calibration, and reclassification.

	30-day mortality			3-year mortality
	Model 1	Model 2	Model 3	Model 1	Model 2	Model 3
Basic demographic/biochemistry	Age, sex, haemoglobin, and creatinine	Age, sex, haemoglobin, and creatinine	Age, sex, haemoglobin, and creatinine	Age, sex, haemoglobin, and creatinine	Age, sex, haemoglobin, and creatinine	Age, sex, haemoglobin, and creatinine
+ Troponin		+ Troponin (positive versus negative)	+ Troponin (positive versus negative)		+ Troponin (positive versus negative)	+ Troponin (positive versus negative)
+ hsCRP			+ hsCRP groups (<2, 2 to 4.9, 5 to 9.9, and 10 to 15)			+ hsCRP groups (<2, 2 to 4.9, 5 to 9.9, and 10 to 15)
Discrimination
AUROC	0.747	0.806	0.812	0.790	0.797	0.803
95% CI	0.745 to 0.750	0.803 to 0.808	0.810 to 0.815	0.787 to 0.793	0.794 to 0.800	0.800 to 0.806
p-Value	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001
p-Value (versus Model 1)	-	<0.001	<0.001	-	<0.001	<0.001
p-Value (versus Model 2)	-	-	<0.001	-	-	<0.001
Reclassification
		versus Model 1	versus Model 2		versus Model 1	versus Model 2
IDI (%)	-	1.1	0.3	-	1.0	0.9
95% CI	-	0.9 to 1.3	0.2 to 0.4	-	0.8 to 1.2	0.7 to 1.1
p-Value	-	<0.001	<0.001	-	<0.001	<0.001
NRI (%)	-	42.6	18.4	-	20.9	13.9
95% CI	-	39.7 to 45.3	14.6 to 21.0	-	19.2 to 22.0	12.7 to 14.9
p-Value	-	<0.001	<0.001	-	<0.001	<0.001

Discrimination of the different models was assessed by comparing the AUROC using the DeLong nonparametric approach. The survIDINRI package on R was used to implement the IDI and NRI for comparing risk prediction models. Likelihood ratio test p-values have been calculated, which confirm the significance of the predictive performance testing using IDI and NRI. The p-values remained significant at <0.001 for all model comparisons.

AUROC, area under the receiver operating characteristic curve; hsCRP, high-sensitivity C-reactive protein; IDI, integrated discrimination improvement; NRI, net reclassification index.

Fig 4 displays the relationship between the negative predictive value of hsCRP and troponin testing with hypothetical mortality. For a potential rule-out test based on risk prediction, there was an improvement in negative predictive values for the lowest hsCRP group (<2 mg/L) in addition to troponin negativity, compared to either test being negative alone, at 30 days and 3 years.

Fig 4

Relationship between negative predictive value of hsCRP and troponin testing with (A) 30-day and (B) 3-year mortality. hsCRP, high-sensitivity C-reactive protein; hsCRP −, negative high-sensitivity C-reactive protein; Tn −, negative troponin.

Discussion

In this large retrospective cohort study, we demonstrate a positive and graded relationship between mildly elevated hsCRP levels (up to 15 mg/L) and all-cause mortality over 3 years. The mortality risk associated with increased hsCRP was independent of age, sex, haemoglobin, WCC, platelet count, creatinine, and troponin positivity, suggesting its value as a clinically useful biomarker in the context of suspected ACS. Compared to a normal hsCRP (<2 mg/L), a mildly raised hsCRP between 10 and 15 mg/L was associated with a 2.2-fold increased mortality risk at 30 days, which persisted at 3 years. In this study, patients who had negative serum troponin but a hsCRP between 10 and 15 mg/L had equivalent mortality to those who were troponin positive but had a normal hsCRP, <2 mg/dL, in Kaplan–Meier analysis at 3 years. Moreover, patients were better reclassified into at-risk mortality groups when hsCRP was included in addition to troponin, suggesting a potential use in this clinical setting.

We limited the cohort by prespecifying strict exclusion criteria that aim to reduce the numbers of patients with infective, inflammatory, or malignant disorders. Therefore, we excluded anyone with abnormal WCC or hsCRP >15 mg/L, with the aim of assembling a cohort that would be informative for clinicians treating patients with suspected ACS. This was possible due to the large population available for study.

The key finding of this study was that within a group of patients with low levels of hsCRP elevation, there was a significant increase in mortality risk in patients presenting with suspected ACS in both the short and long term. In Cox regression analysis with time-dependent covariates, the highest hsCRP group had a HR 2.20 (95% CI 1.88 to 2.57) at baseline, which only marginally attenuated to 2.00 (95% CI 1.75 to 2.28) at 3 years. This long-term risk is clinically significant and may present an opportunity to improve patient selection for novel diagnostics and therapeutics. The striking finding that patients in the highest hsCRP group, but who were troponin negative, had similar mortality to those who were troponin positive but had hsCRP < 2 mg/L, brings into question whether, in clinical practice, we are missing high-risk patients who may benefit from further investigation and treatment [10]. To expand on this, we believe that it is common for patients with a low-normal hsCRP level to be discharged from the emergency department with a caution that they may have a viral infection or nonspecific inflammation, with no comment on mortality risk stratification. Reliance on troponin for risk stratification alone may mean that signals suggesting patient vulnerability are not followed up appropriately. Importantly, the relationships seen with hsCRP and mortality hold when the population is restricted to those with a diagnosis of ACS, rather than suspected ACS.

There has been a very positive response surrounding immunomodulatory drugs in CVD following publication of the CANTOS [15], COLCOT [18], and LoDoCo2 trials [19]. However, CIRT [17] highlighted that not all anti-inflammatory agents would show promise in treating atherosclerosis. Methotrexate in this study had hardly any effect on hsCRP levels, and in tandem, did not reduce CV events, warranting early study termination. Unfortunately, no CRP data from COLCOT or LoDoCo2 are available. Of note, a recent Phase II trial (RESCUE) assessed the efficacy of ziltivekimab, a novel monoclonal antibody against the IL-6 ligand, in patients with hsCRP >2mg/L and chronic kidney disease. This study demonstrated a dose-related reduction in hsCRP (92% with the highest dose), and a large CV outcomes trial is planned [26].

It is difficult in clinical practice to identify the exact time point when an ACS event occurs. The time of diagnosis may not coincide with the time of onset of the ACS event. In order to determine the effect of baseline hsCRP levels rather than hsCRP rises in response to an infarct, we carried out analyses on admission hsCRP levels in comparison to the “peak” troponin level in predicting mortality. For patients with an elevated troponin result, the median number of days between the first admission hsCRP measured and peak troponin level was 3 (IQR 0 to 5) days, infering that the admission hsCRP levels likely represent presentation values rather than the sequelae of the index event.

We excluded individuals who had a CRP >15 mg/L (or abnormal WCC) to limit the population to those without overt infections, malignancies, or systemic inflammatory conditions. It is unlikely that we excluded many patients with ACS due to the low CRP levels measured in patients with ACS in previous studies. In a recent study investigating the association of initial hsCRP levels with MACE and mortality after ACS [27], the median baseline hsCRP level was 10.5 mg/L (IQR, 4.2 to 30.3 mg/L) in ACS patients. On multivariable analysis, higher baseline hsCRP level (HR, 1.36 [95% CI, 1.13 to 1.63]; p = 0.001) was also shown to be independently associated with MACE and mortality.

Recently, in the multicentre matched-control pilot study on CRP apheresis in Acute Myocardial Infarction (CAMI-1), there was a correlation between the CRP gradient between 12 and 32 hours after symptom onset and both myocardial infarct size and the extent of reduction in left ventricular function [28]. The CAMI-1 investigators evaluated the feasibility and safety of CRP apheresis to reduce myocardial infarct size after ST segment elevation myocardial infarction (STEMI) as determined by CV magnetic resonance. While the primary end point of reduction of infarct size was not met, a sufficiently large randomised trial of CRP apheresis therapy in uncomplicated STEMI with the end point of infarct size reduction is currently under development at the Medical University of Innsbruck. This should provide further insight as to whether CRP is also a potentially damaging mediator for CV disease, infarct size, and subsequent mortality risk [29]. However, these results and planned trials need to be viewed in the context of the large mendelian randomisation meta-analysis that concluded that CRP itself [13] (unlike the the IL-6 receptor [30]) is not causally related to coronary heart disease. Indeed, the data from the present study may influence future trial design and lead to careful consideration of the use of hsCRP to select patients at higher CV risk who may derive greater benefit from targeted treatment.

Moreover, an interesting finding in the present study is that those with raised hsCRP and moderate troponin elevations had the greatest mortality risk. While there was a graded relationship between hsCRP level and mortality among patients with different levels of troponin positivity, the mortality risk in the hsCRP groups increased progressively up to >5 to 10×ULN of troponin and then progressively declined in the higher troponin groups. Previous research from our group has demonstrated this inverted U-shaped relation between troponin level and mortality in an unselected cohort of patients undergoing troponin testing. This was primarily due to an interaction between ACS diagnosis and differential receipt of invasive management in this cohort. [21] The >5 to 10×ULN troponin group are likely to represent non-ST segment elevation myocardial infarctions (NSTEMI) than STEMI, and it is perhaps in this population that any targeted anti-inflammatory therapy would have the most clinical benefit.

The observed discrimination and net reclassification when adding hsCRP to our risk models is an important new finding. There have been several studies showing the value of hsCRP as an added covariate for predicting long-term mortality risk in primary prevention populations or those with chronic CVD, leading to developments such as the Reynolds risk score, which improves on the widely adopted Framingham risk score [31]. To our knowledge, there have been no studies that have demonstrated significantly superior AUROC in concert with changes in IDI and NRI that are clinically relevant.

There are some limitations to this study. The present study had access to all-cause, but not specific, mortality data. This limits our ability to understand the causes of death in the patients with the higher hsCRP levels. However, the exclusion of those with abnormal WCC or hsCRP levels >15 mg/L makes it unlikely that sepsis was a major contributor to the excess mortality. Nonetheless, all-cause mortality is clearly a clinically relevant end point. A further limitation is the retrospective nature of the data, where covariates were collated from clinical records, rather than a prospective database that is designed to answer a predefined hypothesis. We were unable to determine the patient comorbidities for our cohort. The median age demographic of patients progressively increases from the lowest to the highest quartile of hsCRP. While we were unable to adjust our analyses by patient comorbidities, all survival analyses were adjusted for age, which may also act as a surrogate marker of patient comorbidities, which are expected to increase in frequency with older age.

The data from this study emphasises the need for further prospective work that explores the inflammation pathway as a therapeutic target to address the still unanswered mortality gap in patients presenting with suspected ACS. We know that recurrent MACE in those undergoing percutaneous coronary intervention (PCI) [32] or presenting with ACS is still unacceptably elevated and shown to be as high as 20% at 3 years [33] despite optimal medical therapy. Our findings indicate a means of selecting those high-risk patients that may benefit from novel therapeutics.

Conclusions

The data from this large real-world population in a modern healthcare system emphasise the importance of hsCRP in this setting and warrant attention and consideration from clinical guideline committees to include hsCRP in risk stratification of patients presenting with suspected ACS.

Disclaimers

The views expressed in this publication are those of the authors and not necessarily those of the NHS, the NIHR or the Department of Health.

STROBE guideline checklist.

STROBE, Strengthening the Reporting of Observational Studies in Epidemiology.

(DOCX)

Click here for additional data file.

Data collection and access plan.

(DOCX)

Click here for additional data file.

Statistical analysis plan.

(DOCX)

Click here for additional data file.

Troponin assays at participating cardiac centres.

(DOCX)

Click here for additional data file.

ICD-10 diagnostic codes used to indicate an ACS diagnosis.

ACS, acute coronary syndrome; ICD, International Classification of Diseases.

(DOCX)

Click here for additional data file.

hsCRP assays at participating cardiac centres.

hsCRP, high-sensitivity C-reactive protein.

(DOCX)

Click here for additional data file.

Adjusted HRs for 3-year mortality according to troponin and hsCRP stratified groups.

HR, hazard ratio; hsCRP, high-sensitivity C-reactive protein.

(DOCX)

Click here for additional data file.

Correlation between hsCRP and troponin level.

hsCRP, high-sensitivity C-reactive protein.

(DOCX)

Click here for additional data file.

Kaplan–Meier mortality curve by hsCRP level and different levels of troponin positivity.

hsCRP, high-sensitivity C-reactive protein.

(DOCX)

Click here for additional data file.

Unadjusted Kaplan–Meier mortality curves by ACS diagnosis.

ACS, acute coronary syndrome.

(DOCX)

Click here for additional data file.

16 Sep 2021

Dear Dr Hartley,

Thank you for submitting your manuscript entitled "High sensitivity C-reactive protein predicts mortality beyond troponin in 102337 patients with suspected acute coronary syndrome" for consideration by PLOS Medicine.

Your manuscript has now been evaluated by the PLOS Medicine editorial staff as well as by an academic editor with relevant expertise and I am writing to let you know that we would like to send your submission out for external peer review.

However, before we can send your manuscript to reviewers, we need you to complete your submission by providing the metadata that is required for full assessment. To this end, please login to Editorial Manager where you will find the paper in the 'Submissions Needing Revisions' folder on your homepage. Please click 'Revise Submission' from the Action Links and complete all additional questions in the submission questionnaire.

Please re-submit your manuscript within two working days, i.e. by Sep 20 2021 11:59PM.

Login to Editorial Manager here: https://www.editorialmanager.com/pmedicine

Once your full submission is complete, your paper will undergo a series of checks in preparation for peer review. Once your manuscript has passed all checks it will be sent out for review.

Feel free to email us at plosmedicine@plos.org if you have any queries relating to your submission.

Kind regards,

Callam Davidson

Senior Editor

PLOS Medicine

4 Nov 2021

Dear Dr. Hartley,

Thank you very much for submitting your manuscript "High sensitivity C-reactive protein predicts mortality beyond troponin in 102337 patients with suspected acute coronary syndrome" (PMEDICINE-D-21-03948R1) for consideration at PLOS Medicine.

Your paper was evaluated by an associate editor and discussed among all the editors here. It was also discussed with an academic editor with relevant expertise, and sent to independent reviewers, including a statistical reviewer. The reviews are appended at the bottom of this email and any accompanying reviewer attachments can be seen via the link below:

[LINK]

In light of these reviews, I am afraid that we will not be able to accept the manuscript for publication in the journal in its current form, but we would like to consider a revised version that addresses the reviewers' and editors' comments. Obviously we cannot make any decision about publication until we have seen the revised manuscript and your response, and we plan to seek re-review by one or more of the reviewers.

In revising the manuscript for further consideration, your revisions should address the specific points made by each reviewer and the editors. Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper. In your rebuttal letter you should indicate your response to the reviewers' and editors' comments, the changes you have made in the manuscript, and include either an excerpt of the revised text or the location (eg: page and line number) where each change can be found. Please submit a clean version of the paper as the main article file; a version with changes marked should be uploaded as a marked up manuscript.

In addition, we request that you upload any figures associated with your paper as individual TIF or EPS files with 300dpi resolution at resubmission; please read our figure guidelines for more information on our requirements: http://journals.plos.org/plosmedicine/s/figures. While revising your submission, please upload your figure files to the PACE digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at PLOSMedicine@plos.org.

We expect to receive your revised manuscript by Nov 25 2021 11:59PM. Please email us (plosmedicine@plos.org) if you have any questions or concerns.

***Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.***

We ask every co-author listed on the manuscript to fill in a contributing author statement, making sure to declare all competing interests. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. If new competing interests are declared later in the revision process, this may also hold up the submission. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT. You can see our competing interests policy here: http://journals.plos.org/plosmedicine/s/competing-interests.

Please use the following link to submit the revised manuscript:

https://www.editorialmanager.com/pmedicine/

Your article can be found in the "Submissions Needing Revision" folder.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Please ensure that the paper adheres to the PLOS Data Availability Policy (see http://journals.plos.org/plosmedicine/s/data-availability), which requires that all data underlying the study's findings be provided in a repository or as Supporting Information. For data residing with a third party, authors are required to provide instructions with contact information for obtaining the data. PLOS journals do not allow statements supported by "data not shown" or "unpublished results." For such statements, authors must provide supporting data or cite public sources that include it.

We look forward to receiving your revised manuscript.

Sincerely,

Callam Davidson,

PLOS Medicine

plosmedicine.org

-----------------------------------------------------------

Requests from the editors:

Please revise your title according to PLOS Medicine's style. Your title must be nondeclarative and not a question. It should begin with main concept if possible. "Effect of" should be used only if causality can be inferred, i.e., for an RCT. Please place the study design ("A randomized controlled trial," "A retrospective study," "A modelling study," etc.) in the subtitle (ie, after a colon).

Please structure your abstract using the PLOS Medicine headings (Background, Methods and Findings, Conclusions).

Abstract Background: Provide the context of why the study is important. The final sentence should clearly state the study question.

Abstract Methods and Findings:

* Please include the important dependent variables that are adjusted for in the analyses.

* In the last sentence of the Abstract Methods and Findings section, please describe the main limitation(s) of the study's methodology.

Financial disclosure: your financial disclosure does not detail the funding source, please provide this information in your response to the submission form. The detailed funding information at the end of the main text should be removed and provided instead as part of your response to the submission form (this will then be published as metadata in the event of publication).

Similar to the above, please remove the ‘Authors’ Contributions’ and ‘Conflict of Interest’ sections from the end of the main text and ensure this information is captured in your submission form (this is essential as your submission form currently states no competing interests exist which is at odds with the main text).

You state in your Data Availability Statement that some restrictions will apply but also then state that all data is contained within the manuscript and/or supporting information files – please update for consistency, bearing in mind our data availability requirements (https://journals.plos.org/plosmedicine/s/data-availability).

At this stage, we ask that you include a short, non-technical Author Summary of your research to make findings accessible to a wide audience that includes both scientists and non-scientists. The Author Summary should immediately follow the Abstract in your revised manuscript. This text is subject to editorial change and should be distinct from the scientific abstract. Please see our author guidelines for more information: https://journals.plos.org/plosmedicine/s/revising-your-manuscript#loc-author-summary

Please update your manuscript to include line numbering in the margin.

Please update your citations to be in square brackets.

Please relocate the information pertaining to the ethical approval from the end of the main text to the methods section.

I could not locate your data acquisition and statistical analysis plan? Please ensure these are included and, when stating that the data acquisition and statistical analysis plan are available in the supplementary material, please reference the relevant section of the supplementary material (e.g. S1).

Please only report P values to 3 decimal places (P<0.001 rather than P<0.0001).

Your study is observational and therefore causality cannot be inferred. Please remove language that implies causality, such as ‘confirms its value as a biomarker’. Refer to associations instead or tone down the language used (e.g. suggests, potential, etc.).

Please add ‘to our knowledge’ when stating that ‘there have been no studies that have demonstrated significantly superior AUROC’ in paragraph 5 of your discussion.

Please remove the One-sentence summary.

References: journal name abbreviations should be those found in the National Center for Biotechnology Information (NCBI) databases.

Please ensure that the study is reported according to the STROBE guideline, and include the completed STROBE checklist as Supporting Information. Please add the following statement, or similar, to the Methods: "This study is reported as per the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guideline (S1 Checklist)."

The STROBE guideline can be found here: http://www.equator-network.org/reporting-guidelines/strobe/

When completing the checklist, please use section and paragraph numbers, rather than page numbers.

Comments from the reviewers:

Reviewer #1: "High sensitivity C-reactive protein predicts mortality beyond troponin in 102337 patients with suspected acute coronary syndrome" studies the association of (mildly) elevated levels of high-sensitivity C-reactive protein (hsCRP) on long-term (3 year) mortality risk in patients with suspected acute coronary syndromes (ACS), independent of troponin level. The main analysis concluded that there was indeed a positive and graded relationship between hsCRP and mortality at baseline and over three years, as summarized in Figures 2B and 3. In particular, stratification by hsCRP level demonstrated increasing 3-year mortality risk given fixed troponin positivity/negativity status (Figure 2B), and increased mortality over the full 3-year period by multivariable time-dependent Cox regression mortality analysis (Figure 3). hsCRP was therefore claimed to be a possible clinically prognostic marker in addition to troponin.

While the primary association between hsCRP and mortality risk appears clear, a number of issues might be considered, largely towards clarifying the methodology:

1. The definition of the upper limit of normal (ULN) for troponin assays might be explicitly stated, and the significance of "99th percentile of ULN" explained, especially since levels of troponin >100x ULN appear possible from Supplementary Figure 2. In particular, if ULN is the threshold on troponin level that separates troponin positive and negative diagnoses, then it would appear more natural to refer to the "99th percentile of troponin values" (to eliminate outliers at the upper end?); this might not be obvious to the more general reader.

2. Moreover, it is stated that the different troponin assays from various academic centres were standardized by scaling results using the ratio of observed troponin value divided by the (unique) ULN for each troponin assay. It might be clarified as to how the ULN for each troponin assay was determined. Is this ULN inherent to the troponin assay model, or empirically calculated on some cohort? Also, were such (scaling) normalizations significant?

3. Supplementary Figure 2 shows sensitivity analyses comparing the mortality curves for different definitions of troponin positivity (1-5x ULN, 5-10x, 10-100x, >100x). It appears that overall mortality risk is generally highest for 5-10x ULN, and actually lower for >100x ULN, than 1-5x ULN, i.e. the association of troponin level with mortality risk is nonlinear (as also briefly raised as an interesting point in the Discussion section). If so, it might be discussed in greater detail as to whether this nonlinear association is expected, and its possible causes.

4. It is stated that the data model involves patients with suspected ACS, as characterized by the request of a troponin test. It might be clarified as to whether this set of patients with troponin test results corresponds to the n=257,948 patients of the initial cohort (Figure 1), and whether the troponin test is exclusively administered towards suspected ACS.

5. It is stated that n=134,517 patients were admitted to hospital and thus had ICD discharge codes, and patients were classified as ACS based on said ICD-10 codes. It might be clarified as to how and where this n=134,517 can be represented in the patient flowchart (Figure 1)

6. It is stated that three statistical risk models were developed for mortality risk prediction (Table 2). The details of these models might be clarified - i.e. are they linear/logistic regression models (and if so, their coefficients might be provided), or other machine learning models, etc. Also, the choice of variables to match for/adjust for (e.g. ethnicity not included) might be briefly justified.

7. For the Kaplan-Meier analysis in Figure 2, it might be explicitly stated if any matching/adjustment was attempted. The "No. at risk" matrices might also be briefly explained/defined.

8. For Figure 3, it might be considered to provide additional time-dependent Cox regression mortality analysis charts stratified on troponin positivity/negativity, for consistency with the previous Kaplan-Meier analysis in Figure 2.

9. Further on Figure 3, the time-dependent analysis would appear to require data from patients at each follow-up interval over the full 3-year period, whereas there does not appear to be a description for the treatment of patients with missing/incomplete data, in Figure 1 and elsewhere. It might be clarified whether all required data was available for all patients involved, or if not, how missing data was handled (this extends to other covariates, other than hsCRP)

10. Figure 4 might be explained in greater detail, with "negative predictive value" in particular being clarified.

11. In the Results section, it is stated that "...Although we tested hsCRP in relation to peak troponin available in predicting mortality, we established that the median number of days between first troponin test and hsCRP was 0 (IQR 0), indicating that our hsCRP and troponin values represent presentation values rather than the sequelae of the index event". The significance of this statement might be clarified further. Does it imply that peak troponin is generally attained on the first test?

12. The claim of "...The independence of hsCRP from all other available covariates including troponin in predicting mortality" in the Discussion section, might be supported with more specific data.

Minor comments:

13. The emphasis on "mildly elevated hsCRP" in describing association with mortality risk might be clarified; from Figure 2, it appears that the association exists over a large range of hsCRP elevation.

14. Assessment of hsCRP independent of troponin level is also variously referred to as "beyond troponin" (in the title) and "above troponin" (in the abstract). It might be considered to standardize the terminology throughout, for clarity.

15. In the Statistical Analysis section, it is stated that "the IDI reflects the change in calculated risk for each individual"; it might be clarified as to whether this change is positive by default (i.e. improvement).

Reviewer #2: This paper describes the incremental long-term prognostic value of high-sensitivity C-reactive protein (hsCRP) above troponin in a large real-world cohort of unselected ACS patients. The objective was to determine if mildly elevated hsCRP was associated with mortality risk, irrespective of troponin level. Eligible patients from 2010 until 2017 were included retrospectively, utilizing NIH Research Health Informatics Collaborative data collected from five hospital centers in Great Britain. Patient selection was based on troponin measurements, normal white blood cell counts and a hsCRP up to 15 mg/L. Patients were followed up until death or censoring within 3 years. To standardize, the ratio of the observed peak troponin value was divided by the ULN for each troponin assay. 102337 patients were included in the analysis. The population was divided into four categories, those with a normal hsCRP <2mg/L, and three groups of incrementally increasing hsCRP (2-4.9 mg/L, 5-9.9 mg/L and 10-15 mg/L). On multivariable Cox regression analysis, there was a positive and graded relationship between hsCRP level and mortality at baseline, which remained at three-years, irrespective of troponin values.

The background for this mega-study is based on the hypothesis that inflammation is a contributory factor for the development of atherosclerosis, supported by RCTs, such as CANTOS, COLCOT and LoDoCo2, the former applying canakinumab and the two latter applying colchicine as anti-inflammatory agents, whereas no effect of methotrexate was obtained in CIRT. These trials were performed in patients with established coronary heart disease, with cardiovascular disease as a primary endpoint, reflecting progression of atherosclerotic disease, whereas the current study presents results related to total mortality, which also includes non-cardiac death, and, ideally, this should have been accounted for when discussing the significance of hsCRP as a marker of atherosclerotic disease.

As patients in the highest as compared to the lowest quartile of hsCRP are 10 years older, one would expect an increasing proportion of co-morbidity through the quartiles of this observational study. These limitations are essential, especially when relating the results to progressive atherosclerotic disease and recommending hsCRP for selection of high-risk patients that may benefit from anti-inflammatory treatment.

Hs-CRP is an acute phase reactant and may not be a reliable indicator of risk, when not measured during steady state. However, the observation that mortality was greatest in patients with raised hs-CRP and moderate troponin elevations, might support the authors' claim.

In short, discussion and limitations require more attention.

Reviewer #3: This manuscript describes the importance of measuring hsCRP in ACS patients as prognostic factor in a large patient cohort. This research question is crucial to further provide information on patient selection for anti-inflammatory therapies and early prognosis of deleterious outcomes.

The main claims are that CRP admission levels of suspected and actual ACS patients significantly correlate with 30-day and 3-year mortality of these patients. It is therefore concluded that CRP should be routinely assessed in ACS patients and not discarded as only a systemic inflammation marker. Together with troponin, risk assessment of ACS patients could be easily improved.

Minor points:

1) The authors should discuss critically and in more detail that all data analysis is based on the initial hsCRP value at hospital admission. They write "that our hsCRP and troponin values represent presentation values rather than sequelae of the index event". The authors should put this in the context of recent publications regarding CRP's active role in the pathophysiology of acute coronary incidents such as AMI and the acute phase response in general. Is there information of the exact time point of CRP measurement and actual diagnosed ACS? This could enlighten whether it is actually the CRP baseline value or already the initial increase of the acute phase response. By excluding patients with CRP > 15 mg/L not only infections but also already ongoing acute coronary events were excluded. The authors also do not mention that in the CAMI-1 trial, a significant association was found between CRP gradient (CRP increase between 12 and 32 hours after symptom onset) and infarct size, LVEF, and circumferential and longitudinal strain. In addition, some publications reported a significant correlation between infarct size and peak CRP, which could also be used as a predictor (most from University of Innsbruck). Both could be discussed in relation to the use of CRP level on admission. In the context of a 2016 meta-analysis showing a significant association between infarct size and mortality (Stone et al.), another point could be discussed.

2) Given the strong evidence for the pathophysiological effect of a high CRP level, the following statement in the manuscript could be supplemented by citing recent literature: "The striking finding that patients in the highest hsCRP group, but who were troponin-negative, had similar mortality to those who were troponin-positive but had hsCRP<2 mg/L, brings into question whether, in clinical practice, we are missing high-risk patients who may benefit from further investigation and treatment."

3) Following this, the authors should also introduce CRP not only as end-protein read-out of the IL-6 cascade. They write this in the context of atherosclerosis and baseline CRP levels but should include CRP's pathophysiological role during and after acute incidents.

4)The statistical methods are well described and mostly appropiate. Among others, the continuous NRI and IDI are used to show the improvement in the prediction model, when hsCRP and troponin positivity are added and p-values are calculated. Continuous NRI and IDI are frequently used, but also prone to error, especially if nested models are used and p-values are calculated (Pepe et al., 2014, Pencina et al., 2015, Burch et al. 2017). The continuous NRI and IDI could be used to describe the improvement of the model, but likelihood-based methods should be used to calculate p-values (Pepe et al. 2013, Burch et al. 2017).

Since there are already significant differences in the AUROC for the different models and there is a clear improvement in NRI and IDI, it is very likely, that the likelihood-based calculation will support the improvement of the models.

Literature:

Pepe et al. 2013: Testing for improvement in prediction model performance, Stat Med. 2013 Apr 30; 32(9): 1467-1482.

Pepe et al. 2014: Net Risk Reclassification P Values: Valid or Misleading?, J Natl Cancer Inst. 2014 Apr; 106(4): dju041.

Pencina et al. 2015: RE: net risk reclassification P Values: valid or misleading?, J Natl Cancer Inst. 2014 Nov 27;107(1):355.

Burch et al. 2017: Net Reclassification Index and Integrated Discrimination Index Are Not Appropriate for Testing Whether a Biomarker Improves Predictive Performance, Toxicol Sci. 2017 Mar; 156(1): 11-13.

Any attachments provided with reviews can be seen via the following link:

[LINK]

1 Dec 2021

Submitted filename: Response to Reveiwer Document.docx

Click here for additional data file.

23 Dec 2021

Dear Dr. Mayet,

Thank you very much for re-submitting your manuscript "The mortality risk prediction of mildly elevated high-sensitivity C-reactive protein beyond troponin in 102337 patients with suspected acute coronary syndrome (NIHR Health Informatics Collaborative CRP-RISK Study): A retrospective cohort study" (PMEDICINE-D-21-03948R2) for review by PLOS Medicine.

I have discussed the paper with my colleagues and the academic editor and it was also seen again by two reviewers. I am pleased to say that provided the remaining editorial and production issues are dealt with we are planning to accept the paper for publication in the journal.

The remaining issues that need to be addressed are listed at the end of this email. Any accompanying reviewer attachments can be seen via the link below. Please take these into account before resubmitting your manuscript:

[LINK]

***Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.***

In revising the manuscript for further consideration here, please ensure you address the specific points made by each reviewer and the editors. In your rebuttal letter you should indicate your response to the reviewers' and editors' comments and the changes you have made in the manuscript. Please submit a clean version of the paper as the main article file. A version with changes marked must also be uploaded as a marked up manuscript file.

Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper. If you haven't already, we ask that you provide a short, non-technical Author Summary of your research to make findings accessible to a wide audience that includes both scientists and non-scientists. The Author Summary should immediately follow the Abstract in your revised manuscript. This text is subject to editorial change and should be distinct from the scientific abstract.

We hope to receive your revised manuscript within 2 weeks. Please email us (plosmedicine@plos.org) if you have any questions or concerns.

We ask every co-author listed on the manuscript to fill in a contributing author statement. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT.

Please ensure that the paper adheres to the PLOS Data Availability Policy (see http://journals.plos.org/plosmedicine/s/data-availability), which requires that all data underlying the study's findings be provided in a repository or as Supporting Information. For data residing with a third party, authors are required to provide instructions with contact information for obtaining the data. PLOS journals do not allow statements supported by "data not shown" or "unpublished results." For such statements, authors must provide supporting data or cite public sources that include it.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Please review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript.

Please note, when your manuscript is accepted, an uncorrected proof of your manuscript will be published online ahead of the final version, unless you've already opted out via the online submission form. If, for any reason, you do not want an earlier version of your manuscript published online or are unsure if you have already indicated as such, please let the journal staff know immediately at plosmedicine@plos.org.

If you have any questions in the meantime, please contact me or the journal staff on plosmedicine@plos.org.

We look forward to receiving the revised manuscript by Jan 06 2022 11:59PM.

Sincerely,

Callam Davidson,

Associate Editor

PLOS Medicine

plosmedicine.org

------------------------------------------------------------

Requests from Editors:

Please update your line numbering such that it is continuous throughout the document (i.e. does not restart with each new page).

Please update your full title to ‘Mortality risk prediction of high sensitivity C-reactive protein in suspected acute coronary syndrome: a cohort study’.

Please trim your abstract as it is currently slightly too long. Key information to include: study design, population and setting, number of participants, years during which the study took place, length of follow up, main outcome measures, quantified main results (with 95% CIs and p values), and important dependent variables adjusted for in the analyses. Additional methodological detail (e.g. detailed exclusion criteria [lines 14-16], standardisation of troponin levels [lines 17-18], exposure classification [lines 18-19]) can be removed. Additionally, you can remove the content from ‘All analyses on hsCRP…’ to ‘…using the peak troponin level’ in paragraph two of the abstract methods and findings (lines 24-27).

Please also condense the ‘limitations’ section of the abstract (lines 10-13) such that it is only a single sentence beginning ‘The main limitations of the study are’, or similar.

Given the observational nature of your study, I suggest updating the first sentence of the abstract conclusions to read ‘These multi-centre, real-world data from a large cohort of patients with suspected ACS suggest mildly elevated hsCRP (up to 15 mg/L) may be a clinically meaningful prognostic marker beyond troponin and point to its potential utility in selecting patients for novel treatments targeting inflammation.’

The Data Availability Statement (DAS) requires revision. If the data are not freely available, please describe briefly the ethical, legal, or contractual restriction that prevents you from sharing it. Please also include an appropriate contact (web or email address) for inquiries (this cannot be a study author).

Please update your Author Summary using the standard PLOS Medicine structure. In brief, summaries should comprise of 2-3 single sentence bullet points for each of the three questions. See full guidance here (https://journals.plos.org/plosmedicine/s/revising-your-manuscript) and consult previous PLOS Medicine articles for examples (https://journals.plos.org/plosmedicine/).

Page 8, line 12: Please cite specific items within the supplementary material rather than citing the entire section (e.g. S1 Analysis Plan). See our website for more detailed guidance (https://journals.plos.org/plosmedicine/s/supporting-information).

Please add the following statement, or similar, to the Methods: "This study is reported as per the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guideline (S1 Checklist)."

Did your study have a prospective protocol? Please state this (either way) early in the Methods section.

Thank you for providing your prospective analysis plan. Changes in the analysis-- including those made in response to peer review comments-- should be identified as such in the Methods section of the paper, with rationale.

Page 18, line 5: Please remove the citation for Supplementary Figure 2.

References: Please ensure all journal names are consistently abbreviated using the abbreviations found in the National Center for Biotechnology Information (NCBI) databases.

Please define the statistical test used to derive the p values in the legend of Table 2.

You allude to statistical testing to determine baseline differences between participants; however no p values are reported in Table 1. Please update Table 1 to include relevant p values and report the statistical test used in the legend.

Thank you for providing your STROBE checklist. Please replace the page numbers with paragraph numbers per section (e.g. "Methods, paragraph 1"), since the page numbers of the final published paper may be different from the page numbers in the current manuscript.

The terms gender and sex are not interchangeable (as discussed in http://www.who.int/gender/whatisgender/en/ ); please use the appropriate term.

Comments from Reviewers:

Reviewer #1: We thank the authors for addressing our previous concerns. However, the previous supplementary figure on Kaplan-Meier curves by hsCRP level and ACS diagnosis appears to have been removed; was there any reason for this?

Reviewer #2: The authors have addressed my concerns. I have no further comments.

Any attachments provided with reviews can be seen via the following link:

[LINK]

10 Jan 2022

Submitted filename: Response to Reviewer Document.docx

Click here for additional data file.

11 Jan 2022

Dear Dr Mayet,

On behalf of my colleagues and the Academic Editor, Dr Sanjay Basu, I am pleased to inform you that we have agreed to publish your manuscript "Mortality risk prediction of high sensitivity C-reactive protein in suspected acute coronary syndrome: a cohort study" (PMEDICINE-D-21-03948R3) in PLOS Medicine.

Before your manuscript can be formally accepted you will need to complete some formatting changes, which you will receive in a follow up email. Please be aware that it may take several days for you to receive this email; during this time no action is required by you. Once you have received these formatting requests, please note that your manuscript will not be scheduled for publication until you have made the required changes.

When making these formatting changes, please also update the following:

* Line 100: Correct the spelling of 'accurately'.

* Line 106: Please update to 'In this study of 102,337 patients with suspected heart attack, a higher hsCRP level was associated with a higher risk of death'.

* Line 113: Please update to 'These findings suggest hsCRP is a clinically meaningful marker of risk of death in addition to troponin in patients with suspected heart attack'.

In the meantime, please log into Editorial Manager at http://www.editorialmanager.com/pmedicine/, click the "Update My Information" link at the top of the page, and update your user information to ensure an efficient production process.

PRESS

We frequently collaborate with press offices. If your institution or institutions have a press office, please notify them about your upcoming paper at this point, to enable them to help maximise its impact. If the press office is planning to promote your findings, we would be grateful if they could coordinate with medicinepress@plos.org. If you have not yet opted out of the early version process, we ask that you notify us immediately of any press plans so that we may do so on your behalf.

We also ask that you take this opportunity to read our Embargo Policy regarding the discussion, promotion and media coverage of work that is yet to be published by PLOS. As your manuscript is not yet published, it is bound by the conditions of our Embargo Policy. Please be aware that this policy is in place both to ensure that any press coverage of your article is fully substantiated and to provide a direct link between such coverage and the published work. For full details of our Embargo Policy, please visit http://www.plos.org/about/media-inquiries/embargo-policy/.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Thank you again for submitting to PLOS Medicine. We look forward to publishing your paper.

Sincerely,

Callam Davidson

Associate Editor

PLOS Medicine