The C-reactive protein Albumin ratio was not consistently associated with cardiovascular and all-cause mortality in two community-based cohorts of 70-year-olds

Abstract

C-reactive protein (CRP)/Albumin ratio (CAR) seems to mirror disease severity and prognosis in several acute disorders particularly in elderly patients, which we aimed to study. As method we use a prospective study design; the Prospective Investigation of the Vasculature in Uppsala Seniors (PIVUS; n = 912, women 50%; mean age 70 years, baseline 2001 and 2004, median follow-up 15.0 years, end of follow-up 2019) and the Uppsala Longitudinal Study of Adult Men (ULSAM, n = 924 mean age 71 years, baseline 1991–1995, median follow-up 15.6 years, end of follow-up 2016). Serum samples were used for analyses of CRP and Albumin. Cox regression analyses were performed for cardiovascular and all-cause mortality in models adjusting for several factors (age; physical activity; Interleukin-6; cardiovascular (CVD) risk factors: smoking, BMI level, systolic blood pressure, LDL-cholesterol, and diabetes), with 95% confidence interval (CI). When adjusting for age and CVD risk factors, CAR was significantly associated with cardiovascular mortality for meta-analyzed results from PIVUS and ULSAM, HR 1.09 (95% 1.01–1.18), but neither in PIVUS (HR 1.14, 95% CI 0.99–1.31) nor in ULSAM (1.07, 95% CI 0.98–1.17). Additionally, CAR was significantly associated with all-cause mortality in ULSAM 1.31 (95% CI 1.12–1.54) but not in PIVUS HRs 1.01 (95% 0.089-1.15). The predictive value of CAR was similar to CRP alone in PIVUS and ULSAM and slightly better than albumin for the prediction of CVD-mortality in ULSAM. In conclusion, CAR was not consistently associated with cardiovascular and all-cause mortality in the two cohorts. The prognostic value of CAR for long-term CVD-mortality was similar to CRP.

Keywords:

• Cardiovascular mortality
• diabetes
• blood pressure
• cRP
• albumin

Introduction

CRP is a well-known acute phase protein (APP) reflecting inflammation. CRP alterations have been observed in different malignant diseases, acute inflammatory diseases, acute infections, and trauma, and been associated with mortality [1]. Additionally chronic diseases such as diabetes, atherosclerosis, heart failure, stroke dementia as well as some other neurogenerative disorders are associated with higher CRP levels [2,3]. Finally, elevated CRP levels has also been associated with different stages of frailty and CRP is regarded as a marker of frailty [4]. Besides, CRP levels correlate with BMI, probably as the adipose tissue produces a large number of proinflammatory cytokines [5].

Albumin is also an acute phase reactant. Besides, the physiological role of albumin is to maintain normal osmotic colloid pressure, and to act in intravascular transport of molecules. Additionally, it is also of importance in lipid metabolism as well as thrombosis hemostasis. Low circulating albumin (hypoalbuminemia) has been associated with the severity of several diseases such as malignant diseases, chronic inflammatory diseases, diabetes mellitus acute diseases, sepsis [6] and also cardiovascular disease [7]. Albumin is, in addition to its role to mirror states of inflammation, also considered as a biomarker of malnutrition and poor health status [8], and is negatively associated with both frailty and sarcopenia [4].

Recent studies have reported an emerging role for the C-reactive protein (CRP) albumin ratio (CAR) in acute ill patients and particularly geriatric patients. The CAR was initially proposed as a mortality-prediction tool in acute medical ill patients [9], however data suggest that the CAR could be applicable for other state of diseases. During the search for prognostic biomarkers for outcome in COVID-19 infection, several groups found the CAR to have a specific sensitivity to predict outcome in acute ill geriatric patients [10]. Furthermore, several groups have found a predictive value for outcome in different malignant diseases as well as in infectious diseases specifically for geriatric patients [11–14]. Importantly, the CAR also predicted hospital mortality in a cohort of unsorted geriatric patients admitted to the emergency department (ED) [10,15].

Taken together, CAR seems to be of significant value in the clinical setting with acutely ill patients. The role of CAR in the general elderly population is less well studied. Therefore, our main aim was to analyze CAR and to study the association between CAR and long-term cardiovascular and all-cause mortality in two Swedish cohorts of 70-year-olds. Based on previous studies in the field, we hypothesized that higher CAR would be associated with an increased mortality risk.

Methods

Study samples

The Prospective Investigation of the Vasculature in Uppsala Seniors (PIVUS)

All 70-year old men and women living in Uppsala, Sweden, between 2001 and 2004 were eligible for the PIVUS study (described in detail on http://www.medsci.uu.se/pivus/pivus.htm) [16]. Of 2025 invited individuals, 1016 agreed to participate, and 912 were included in the analyses (with 104 excluded owing to missing values).

The Uppsala Longitudinal Study of Adult Men (ULSAM)

The ULSAM study was initiated in 1970. All 50-year-old men, born in 1920–1924 and living in Uppsala, Sweden, were invited to a health survey, focusing at identifying cardiovascular risk factors (described in detail on http://www.pubcare.uu.se/ULSAM) [17]. The present study used the third examination cycle as baseline, when participants were approximately 70 years old. Of 1681 invited individuals, 1221 agreed to participate, and 924 were included in the analyses (with 297 excluded owing to missing data).

All participants in PIVUS and ULSAM gave written informed consent, and the Ethics Committee of Uppsala University approved the study protocols. Both studies were conducted according to the Declaration of Helsinki.

Baseline investigations

The investigations in PIVUS and ULSAM were performed using similar standardized methods, including anthropometrical measurements, blood pressure, blood sampling, and questionnaires regarding socioeconomic status, medical history, smoking habits, medication, and physical activity level [16,17]. Venous blood samples were drawn in the morning after an overnight fast and stored at −70 °C until analysis.

High-sensitive CRP measurements were performed by latex-enhanced reagent (Siemens) with the use of a BN ProSpec^® analyzer (Siemens). Serum albumin was measured with spectrophotometry using bromine cresol green and reagents from Boehringer Mannheim [18]. The instrument used was Hitachi 717 or 911 from Hitachi, Japan. Diabetes mellitus was diagnosed as fasting plasma glucose ≥7.0 mmol/L (≥126mg/dL), or use of anti-diabetic medication.

Statistical analysis

Longitudinal analyses

Multivariable Cox regression models were used to calculate hazard ratios for long-term cardio-vascular disease (CVD) and all-cause mortality from 70 years of age and onwards in both cohorts. The following multivariable models were used:

• Model 1 crude to show the unadjusted association between CAR levels and events, CVD and all-cause mortality;

• Model 2 adjusted for age to show that the fact that older individuals have an increased CAR level as well as a higher event rate;

• Model 3 adjusted for age, sex (PIVUS) and BMI level

• Model 4 adjusted for age and physical activity to see if physical activity level can explain the association between CAR and events;

• Model 5 adjusted for age and inflammatory markers to see if CAR has a role beyond other markers of inflammation

• Model 6 adjusted for established cardiovascular risk factors (smoking, BMI level, systolic blood pressure, LDL-cholesterol, and diabetes) to see if CAR is an independent risk factor.

We also compared Harrell’s C statistics for CVD-mortality between CAR, CRP alone as well as albumin alone, to determine if the model discrimination of CAR was improved. The statistical software package STATA 15.0 (StataCorp, College Station, TX) was used.

We used the statistical significance level of p < 0.05 and regarded findings that were significant at that level in both cohorts. We used several models, but as the exposures and the outcome is fixed in all models for CAR, the tests cannot be deemed independent.

Results

Baseline data for the two included cohorts are shown in Table 1.

Table 1. Baseline characteristics of the participants in the included cohorts.

	PIVUS	ULSAM
	(n = 912)	(n = 924)
Age, years, mean (SD)	70.2 (0.15)	71.0 (0.62)
Sex, female, %	50	0
BMI, kg/m^2, %:
<18.5	0.88	0.43
18.5–24.9	32.2	36.8
25–29.9	45.0	51.5
≥30	21.9	11.3
Systolic BP, mm HG (SD)	150 (22.5)	133 (16.0)
LDL-cholesterol, mmol/L (SD)	3.40 (0.87)	3.90 (0.89)
Smoking status, %:
Non-smoker	48.0	80.7
Former smoker	41.2	NA
Current smoker	10.8	19.3
Diabetes, %	11.3	14.3
Physical activity level:
METs/week (SD)	259 (4.84)	NA
sedentary, %		3.14
moderate, %		33.7
regular, %		56.9
athletic, %		6.30
S-Albumin, g/L (SD)	40.3 (2.68)	43.6 (2.60)
S-C-reactive protein, mg/L (SD)	2.07 (2.75)	2.78 (2.79)
Interleukin-6 (IL-6), ng/L (SD)	1.84 (1.56)	NA
Excluded in analysis	66	0

Results regarding CVD-mortality are shown in Table 2, and all-cause mortality in Table 3, for PIVUS, and ULSAM using different models. For CVD-mortality, in most models, the HRs were significant, except for Model 6, which was adjusted for smoking, BMI level, systolic blood pressure, LDL-cholesterol, and diabetes. The HRs ranged between 1.14–1.19 in PIVUS, and 1.07–1.13 in ULSAM. We also performed a meta-analysis of model 6, i.e. with CVD risk factors, with HR 1.09 (95% CI 1.01–1.18).

Table 2. The relative risk of cardio-vascular mortality shown as hazard ratios (HRs) for CRP/Albumin ratio (CAR) in the PIVUS cohort (n = 846 observations) and the ULSAM cohort (n = 924 observations).

	PIVUS			ULSAM
	HR	95% CI	p>|z|	HR	95% CI	p>|z|
Model 1	1.19	1.04–1.36	0.0095	1.13	1.04–1.24	0.0064
Model 2	1.19	1.04–1.36	0.0096	1.13	1.03–1.23	0.0079
Model 3	1.18	1.02–1.35	0.0215	1.10	1.01–1.21	0.0301
Model 4	1.18	1.03–1.35	0.0162	1.12	1.03–1.23	0.0104
Model 5	1.17	1.02–1.34	0.0263
Model 6	1.14	0.99–1.31	0.063	1.07	0.98–1.17	0.14

Model 1 crude; Model 2 adjusted for age; Model 3 adjusted for age, sex (PIVUS) and BMI level; Model 4 adjusted for age and physical activity; Model 5 adjusted for age and inflammatory markers (PIVUS); and Model 6 adjusted for BMI level, smoking, systolic blood pressure, LDL-cholesterol, and diabetes.

Table 3. The relative risk of all-cause mortality shown as hazard ratios (HRs) for CRP/Albumin ratio (CAR) in the PIVUS cohort (n = 846 observations) and the ULSAM cohort (n = 924 observations).

	PIVUS			ULSAM
	HR	95% CI	p>|z|	HR	95% CI	p>|z|
Model 1	1.02	0.90–1.16	0.736	1.36	1.16–1.59	0.0001
Model 2	1.02	0.90–1.16	0.737	1.34	1.15–1.57	0.0002
Model 3	1.04	0.91–1.19	0.519	1.33	1.14–1.56	0.0003
Model 4	1.01	0.89–1.15	0.905	1.31	1.12–1.54	0.0007
Model 5	1.01	0.89–1.16	0.829
Model 6	1.01	0.89–1.15	0.919	1.31	1.12–1.54	0.0007

Model 1 crude; Model 2 adjusted for age; Model 3 adjusted for age, sex (PIVUS) and BMI level; Model 4 adjusted for age and physical activity; Model 5 adjusted for age and inflammatory markers (PIVUS); Model 6 adjusted for smoking, BMI level, systolic blood pressure, LDL-cholesterol, and diabetes.

For all-cause mortality, the HRs were close to 1.0 in all models for PIVUS (1.01–1.04), but statistically significant for ULSAM with HRs ranging 1.31–1.36.

Finally, we tested Harrell’s C statistics and if there were any differences in the prediction of CVD-mortality models between CAR and CRP and albumin alone (Table 4). There was no significant difference in Harrell’s C between CAR and CRP in the cohorts. CAR had slightly higher Harrell’s C than albumin in ULSAM but not in PIVUS.

Table 4. Harrell’s C-statistics for CRP-Albumin ratio (CAR) compared to CRP and Albumin alone, respectively, with cardio-vascular mortality as outcome in model adjusting for age and CVD risk factors; hazard ratios comparing Harrell’s C between the different biomarkers.

	CAR	CRP	Albumin	CAR-CRP	CAR-albumin
	Coeff (95% CI)	Coeff (95% CI)	Coeff (95% CI)	HR	HR
CVD-mortality
PIVUS	0.596 (0.540–0.652)	0.594 (0.538–0.651)	0.596 (0.539–0.654)	1.00 (1.00–1.01)	1.00 (0.99–1.01)
Adjusted for CVD-factors
ULSAM
Adjusted for CVD-factors	0.611 (0.573–0.649)	0.613 (0.575–0.650)	0.600 (0.561–0.638)	1.00 (0.99–1.01)	1.01 (1.00–1.02)

CVD-factors: smoking, BMI level, systolic blood pressure, LDL-cholesterol, and diabetes.

Discussion

Main findings

We found conflicting results regarding association between the CAR and both CVD-mortality and all-cause mortality in the studied community-based cohorts of healthy elderly subjects. CAR showed no higher predictive value than CRP when tested by Harrell’s C, and only slightly higher when compared to albumin in one of the two cohorts. The clinical importance of analyzing the CAR in the general aged population seems to be of limited value but should be studied further.

Comparisons with other studies

Exposure of acute illness such as acute inflammation/infection or trauma induces an acute-phase response in the host. This acute response involves different pathophysiological mechanisms such as for example fever, leukocytosis as well as stimulation of hepatic production of acute phase proteins (APP) by cytokines. APP’s are normally divided into negative and positive APP. Negative APP’s are for example albumin and transferrin and positive APP’s are for example CRP, thrombin, and d-dimer protein [19].

The CAR has been studied in several acute diseases including unsorted patients at emergency departments (ED) [10,15], critically ill patients [3], sepsis [14,20], stroke [21] myocardial infarction [22] and respiratory diseases such as acute COVID-19 infection [23] and pneumonia [24]. Moreover, the CAR has been found to be of prognostic value also in chronic diseases such as advanced malignant diseases [11–13,25], pulmonary arterial hypertension [26] and hypertension [23]. Finally, the CAR has been suggested to mirror frailty and malnutrition in elderly patients [27,28].

Mechanisms

A strong association between CRP and mortality has been shown earlier in this cohort [29] as well as in other cohorts [1]. Our findings confirm these earlier studies and the strong value of CRP as a prognostic marker for mortality in the general population.

Circulating albumin most likely reflects both inflammatory and nutritional status. Some studies, specifically addressing patients with renal impairment, suggest no association between albumin and CVD after adjustments for CRP [30–32]. Interestingly, in a cohort of 70-year-old community dwelling individuals, an association between low concentrations of serum albumin and all-cause mortality was found [33]. In contrast, no association between low serum albumin and CVD or cancer mortality was found [33]. Furthermore, the same group found a significant association between low serum albumin and CVD mortality as well as all-cause mortality in a cohort of 80-year-old community dwelling individuals [34]. Taken together, the data in the literature suggest a weaker association between low serum albumin and specifically CVD comorbidity in the general population. Moreover, CVD mortality represents a significant part of the total mortality in this cohort [29]. Thus, we believe that the most likely explanation for the lower predictive value for albumin and total mortality in this cohort is the large number of CVD deaths found in in the present study.

The CAR considers both the inflammatory as well as the nutritional state of the patients and has been strongly associated with mortality in acute ill patients with high inflammatory response and poor nutritional state and has been found of great use in the elderly population. We used two community-based cohorts and individuals within these are most likely in general in a good physical and nutritional state. Hence, large variations in the inflammatory state are not expected. Similarly, during a disease process, CRP typically increases and albumin decreases, but in early phases, albumin will remain on a normal level in otherwise healthy patients. In malnutrition, as seen in frailty or sarcopenia, serum albumin is lowered and indicates a higher risk for the patient. We believe, however, that the individuals in the included cohorts that we studied, were probably not in advanced stages of frailty or sarcopenia [4].

The role of the CAR in the general aged population has been less well studied. To our knowledge, only a few groups have addressed this question. The predictive role for the CAR and the development for SIRS after surgery has been studied in a general population [35], and a significant association was found between increased CAR and development of postoperative SIRS. The association between CAR and mortality was not studied.

The association between CAR and total mortality shown here, even though with a similar magnitude as for CRP alone, suggests a possible predictive role for the CAR in the general elderly population. The lack of large perturbations in albumin and CRP is most likely due to the relatively healthy subjects in the cohorts we studied and we thus believe that CAR has a greater predictive value in cohorts of more advanced ages and increased frailty. It would also be of interest to study high risk cohorts. Another question is whether sex differences could affect the risk. However, when adjusting for sex in the PIVUS cohort only marginal differences were found.

Clinical implications

In this study of men and women being 70 years of age at baseline, the predictive ability of CAR was similar to CRP. The inflammatory elements seem to be more important in these relatively healthy participants invited from the general population. Yet, with increased age and frailty, further studies regarding the potential of CAR remains to be studied in detail in larger cohorts with good data on frailty and with markers of nutritional status.

Strengths and limitations

The present study has several strengths. First, to minimize the risk of chance findings, we used two independent community-based study samples with longitudinal data on mortality. Second, National Swedish Registers are shown to have a high coverage [36]. Third, the detailed characterization of the study participants enabled us to control for a large number of established risk factors. Fourth, to our knowledge, the ULSAM and PIVUS cohorts are the largest community-based cohorts of elderly that have analyzed CAR and with follow-up data on mortality so far.

Limitations include the unknown generalizability to other age- and ethnic groups. Thus, additional large-scale studies in other ethnicities and in other age groups are needed to properly validate our findings and to establish optimal thresholds, in order to identify individuals at an increased risk.

Conclusion

CAR was not consistently associated with cardiovascular and all-cause mortality in two independent cohorts of elderly. The prognostic value of CAR for long-term CVD-mortality was similar as CRP alone. Thus, our data does not provide additional support for using CAR for mortality prediction in elderly in clinical practice.

Authors’ contribution

PW proposed the study and contributed to the interpretation of the results and manuscript drafting. ACC contributed to the interpretation of the results and manuscript writing. AL contributed to the interpretation of the results and manuscript writing. JH had access to the databases, and contributed to the interpretation of the results and revision of the manuscript. TF analyzed the data, and contributed to the interpretation of the results and revision of the manuscript. TR contributed in overall scientific coordination and contributed to the interpretation of the results and manuscript writing.

Ethical approval

All methods were carried out in accordance with relevant guidelines and regulations. Both the PIVUS and the ULSAM studies have ethical approvals from Regionala Etikprövningsnämnden EPN, Uppsala, Sweden.

Informed consent

Informed consent was obtained from all subjects and/or participants. This manuscript contains no individual person’s data, and therefore can be publish without consent from the participants.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

Both the PIVUS and the ULSAM studies are listed in the SND (Swedish National Data Service). Access to data is limited, contact persons are for PIVUS Lars Lind ([email protected]) and for ULSAM Lars Lannfelt ([email protected]).