Serum bicarbonate as a cardiovascular risk factor: evolving from ‘non-traditional’?

Cardiovascular disease (CVD) remains the leading cause of premature death globally [1], and identifying modifiable CVD risk factors continue to be an important goal. Individuals with chronic kidney disease (CKD) suffer premature death more so than those without CKD [2], and do so more commonly than developing end-stage kidney disease [3]. This enhanced CKD-related mortality is due predominantly to increased risk for, and earlier onset of, CVD [2, 3] including ischemic heart disease [4] and cerebrovascular accident or stroke [5], the two top global CVD contributors to premature mortality [1]. Consequently, CKD is an attractive model to study in order to identify CVD risk contributors that might be present disproportionately, but not exclusively, in CKD patients that when addressed might reduce the risk for subsequent CVD. Metabolic acidosis, reflected by reduced serum bicarbonate ([HCO3]) concentration, further increases CVD and overall mortality risk in nontransplant CKD [6] and increases the risk for mortality [7] and ischemic CVD events [8] in kidney transplant recipients. Furthermore, the risk for mortality and adverse CVD outcomes increases in patients with CKD as serum [HCO3] decreases due to metabolic acidosis and this increased risk extends into the normal range for serum [HCO3] [9]. The association of low serum [HCO3] with an increased mortality is not limited to individuals with CKD but includes general populations as well [10, 11]. Indeed, high acid-producing diets in general populations, most of whom presumably do not have CKD, are associated with increased overall and CVD mortality [12]. Although robust mechanisms help protect against reductions in serum [HCO3] in response to acid-producing diets and other acid challenges to acid–base status, acid-producing diets can cause small but measurable reductions in serum [HCO3], typically within the range of normal for clinical laboratories [13–15]. Together, these data support that an acid milieu reflected by reduced serum [HCO3] is associated with increased CVD risk. Nevertheless, the association between low serum [HCO3] and increased CVD risk has not been a consistent finding. Other investigators did not show an association of low serum [HCO3] with CVD risk in patients with CKD due to diabetes [16] or in healthy older individuals [10]. Further studies must better clarify if low serum [HCO3] indeed predicts increased CVD risk and if so, if the association is limited to particular populations. The study of Dobre et al. [17] in this issue helps shed further light on the potential association between low serum [HCO3] and increased CVD risk, including populations for which this association might be most applicable. These investigators examined the association between serum [HCO3] and CVD events in a secondary analysis of the Systolic Blood Pressure Intervention Trial (SPRINT), whose primary analysis was to determine the effect of intensive blood pressure (BP) control on CVD morbidity and mortality in hypertensive individuals with and without CKD but without diabetes. This database importantly allowed investigators to determine if intensive BP control was more beneficial in individuals with low compared with normal baseline serum [HCO3] and in individuals with and without CKD. SPRINT was a multicenter, randomized clinical trial that examined if intensive systolic BP control (<120 mmHg) reduced cardiovascular morbidity/mortality compared with standard systolic BP control (<140 mmHg). The investigators recruited 9361 USA participants, 9334 (99.7%) of whom had baseline serum [HCO3], aged ≥50 years old with hypertension, and followed them for a median of 3.33 years. Eligible participants had systolic BP 130–180 mmHg, and at least one of the following four cardiovascular risk factors: clinical or subclinical CVD, estimated glomerular filtration rate (eGFR) 20–60 mL/min/1.73 m2, Framingham 10-year risk score ≥15% or age ≥75 years. Exclusion criteria included diabetes mellitus, history of stroke, cardiovascular event or procedure in the prior 3 months, symptomatic heart failure in the past 6 months, left ventricular ejection fraction <35%, proteinuria >1 g/day or the equivalent based on spot urine measurement, polycystic kidney disease, recent immunosuppression, coexisting disease likely to affect survival and organ transplant. The investigators measured serum [HCO3] only at baseline. The SPRINT primary outcome was a composite of adjudicated myocardial infarction (MI), acute coronary syndrome not resulting in MI, stroke, acute decompensated heart failure or death from cardiovascular causes. Secondary outcomes included death from any cause, acute decompensated heart failure and two kidney outcomes. The primary kidney outcome for participants with CKD at baseline (defined as eGFR <60 mL/min/1.73 m2) was the time to the first occurrence of eGFR reduction ≥50%, long-term dialysis or kidney transplantation. The secondary kidney outcome, for participants with eGFR ≥60 mL/min/1.73 m2 at baseline, was the time to the first occurrence of eGFR reduction ≥30% to an eGFR <60 mL/min/1.73 m2. The investigators defined incident microalbuminuria as an increase in urine albumin to creatinine ratio (UACR) to ≥30 mg/g for participants with UACR <30 mg/g at baseline. They defined incident macroalbuminuria as an increase in UACR to ≥300 mg/g for participants with UACR <300 mg/g at baseline. They obtained demographic and clinical information by questionnaires, interviews and physical examinations. The investigators defined the history of CVD through clinical evidence of coronary artery disease or peripheral arterial disease. Population demographics were reasonably representative of individuals followed in most primary care and nephrology outpatient settings. The mean age was 67.9 years, 35.5% were women, 29.9% were African American and 20.1% had a history of CVD. Mean baseline serum [HCO3] was 26.3 mEq/L with 3.4% having levels <22 mEq/L, meeting the general criterion for metabolic acidosis [18]. Mean baseline eGFR was 72 mL/min/1.73 m2, with 27.3% of participants having eGFR <60 mL/min/1.73 m2. Participants with serum [HCO3] <22 mEq/L were more likely to be younger, current smokers, have a history of CVD, lower eGFR and higher albuminuria, and to be taking Angiotensin converting Enzyme inhibrition/Angiotensin receptor blocker (ACEi/ARB), and were less likely to be taking diuretics. During a median follow-up of 3.33 years, 6.6% of participants experienced a primary CVD event. After multivariate adjustment including eGFR and albuminuria, baseline serum [HCO3] <22 mEq/L was associated with a 54% higher risk for the primary CVD outcome compared with the reference group. Nonlinear analyses were generally congruent with these categorical results, indicating higher CVD risk for lower serum [HCO3]. Subgroup analysis showed that the association between serum [HCO3] and CVD outcomes was similar between participants with and without CKD, and there was no indication of heterogeneity in the effect of intensive BP control among the serum [HCO3] strata. Specifically, low serum [HCO3] was associated with increased CVD risk in participants with and without CKD, and the benefit of intensive BP control was not different between participants with low compared with normal serum [HCO3]. During a median follow up of 3.36 years, 4.1% of participants died. Participants with low serum [HCO3] had a higher incidence of all-cause mortality (2.3%/year) compared with those with normal serum [HCO3] (1.3%/year). After multivariate adjustment, participants with low serum [HCO3] did not have higher risk of all-cause mortality compared with the reference group. Nonlinear analysis showed a U-shaped association between serum [HCO3] and all-cause mortality with the lowest mortality risk associated with serum [HCO3] near 28 mEq/L, a value considerably great than 22 mEq/L, which is commonly accepted as the point at which patients are deemed to have metabolic acidosis. This increased mortality for lower serum [HCO3] extending into serum [HCO3] within the normal range was reported previously [9]. Intensive BP lowering was associated with reduced all-cause mortality irrespective of serum [HCO3]. Lower serum [HCO3] was associated with higher all-cause mortality in participants with CKD than in those without CKD. During median follow-up of 3.33 years, 1.9% of participants developed heart failure events. Serum [HCO3] was not associated with heart failure events in adjusted analyses or when analyzing serum [HCO3] as a categorical variable. Subgroup analysis showed that the association between serum [HCO3] and heart failure was similar between participants with and those without CKD. The report by Dobre et al. [17] contributes importantly to the evolving story as to a potential association between low serum [HCO3] and CVD risk. First, it adds to published data supporting the association across individuals with and without CKD, supporting low serum [HCO3] as a more general CVD risk predictor and not limited to predicting CVD risk in only patients with CKD. Second, it emphasizes the better predictive value of low serum [HCO3] for CVD outcomes more directly related to ischemia like MI and stroke, and less so for those like heart failure that might be mediated additionally by nonischemic factors. Third, it shows the CVD benefits of intensive BP lowering irrespective of participant baseline serum [HCO3], thereby supporting intensive BP lowering across participant baseline serum [HCO3]. Fourth, it affirms earlier reports [9] that the predictive power of low serum [HCO3] is not confined to individuals with metabolic acidosis (serum [HCO3] <22 mEq/L) but includes serum [HCO3] that extends into the normal range. Relatedly, the data support that the serum [HCO3] that is associated with the lowest CVD risk was 28 mEq/L, a value that today would be considered to be at the higher end of the normal range. Fifth, this report supports the association of low serum [HCO3] and increased CVD risk in individuals without diabetes. This important study by Dobre et al. [17] has many strengths including a large number of randomized, nondiabetic individuals with a good ethnic mix who are at CVD risk, and a large number of participants with CKD, all of which allowed for a thorough analysis of a potential link between serum [HCO3] and CVD outcomes. It has the weakness of all observational studies that raise the question of causality. One major weakness is having only baseline, and not follow-up, serum [HCO3], reflecting that this was not a primary analysis of the study. As with all such studies using serum [HCO3] as a proxy for metabolic acidosis, we do not have the necessary additional objective and history/physical exam data to confirm that low serum [HCO3] was indeed due to metabolic acidosis and not respiratory alkalosis, or was part of a mixed acid–base presentation. Ideally, we also would like to have had longer follow-up. The observation by Dobre et al. [17] and others [10, 11] that low serum [HCO3] is associated with increased CVD risk in individuals with and without CKD supports the overall importance of an acid milieu increasing CVD risk. Diets of developed societies are acid-producing because of the preponderance of acid-producing animal-sourced foods compared with base-producing plant-sourced foods [19], such that individuals living in these societies are undergoing a constant acid challenge. Individuals given increments of dietary acid cumulatively excrete less acid than ingested [20], consistent with sustained acid accumulation that is greater in individuals with reduced compared with normal eGFR [21]. Such acid accumulation can be substantial with acid-producing diets as eGFR decreases over time, which yields small but measurable decreases in serum [HCO3] and increases in the anion gap [15, 16]. The observation that the increased CVD risk associated with low serum [HCO3] extends into the normal range for serum [HCO3] supports that acid accumulation increases CVD risk without necessarily causing metabolic acidosis as currently defined [18]. Acid accumulation related to the acid-producing diets of developed societies might increase CVD risk, particularly due to ischemia, by a number of potential mechanisms. Signal transduction pathways that lead to increased acid excretion intended to reduce acid accumulation can overlap pathways that activate vascular endothelial cell inflammation [22], an important contributor to CVD [23]. For example, the acid accumulation that increases angiotensin II expression, an important mediator of increased kidney acid excretion [24], enhances vascular inflammation, fibrosis and extracellular matrix formation [25]. Thus, responses to acid accumulation can cause vascular endothelial cell inflammation, thereby increasing CVD risk. Furthermore, interleukin 10, which works to reduce the effects of proinflammatory cytokines [26], increases in serum level, reflecting increased inflammation, as eGFR declines [27], and its serum level is directly associated with increased CVD risk in patients with CKD [27]. Oxidative stress, defined as a state of imbalance between excessive oxidant (free) radicals and insufficient degradation of those radicals by endogenous antioxidant systems, contributes to vascular injury, which mediates CVD [23]. A component of the response to acid accumulation is to sequester and buffer it intracellularly. The resulting increased intracellular acidity releases iron bound to intracellular protein, thereby forming hydroxyl radicals that cause tissue-damaging oxidative stress [28], likely contributing to oxidative stress associated with metabolic acidosis in CKD [29]. Sustained high serum and local levels of endothelin contribute vascular pathology that leads to adverse CVD outcomes [30]. Acid accumulation, even when insufficient to cause metabolic acidosis, increases serum levels and urine excretion of endothelin-1, both of which were reduced by oral alkali [21]. Deciphering mechanisms as to how an acid milieu reflected by reduced serum [HCO3] appears to increase CVD risk in individuals with and without CKD will help design interventions to reduce CVD outcomes, with their devastating consequences. The report by Dobre et al. [17] adds to the growing body of evidence that should encourage continued basic studies to determine contributing mechanisms and encourage interventional studies to determine if reducing this milieu is an effective adjunct to traditional CVD risk reduction strategies. Interventions to ameliorate this acid milieu include low acid/high base-producing diets (high in base-producing foods like fruits and vegetables and low in acid-producing food components like animal-sourced foods) and/or oral alkalies like sodium bicarbonate and sodium citrate to neutralize the accumulated acid. Such studies will also help determine if low serum [HCO3] can be moved from ‘non-traditional’ to ‘traditional’ CVD risk factors. D.E.W. receives a portion of his salary paid through his employing institution to serve as a consultant for Tricida, Inc. (San Francisco) and as Principle Investigator on NIH grant #R21DK113440. (See related article by Dobre et al. Serum bicarbonate and cardiovascular events in hypertensive adults: results from the Systolic Blood Pressure Intervention Trial. Nephrol Dial Transplant 2020; 35: 1377–1384)