Discovery and validation of agonistic angiotensin receptor autoantibodies as biomarkers of adverse outcomes

INTRODUCTION

The Renin-Angiotensin System (RAS) has been identified in subcellular organelles such as the nucleus and mitochondria, in organisms without a pressurized vasculature, and in virtually every human organ system.^{1, 2} An independently regulated RAS in human monocytes has been identified.³ Beyond its classic role in homeostatic regulation of blood pressure and fluid balance, the RAS has been implicated in tissue remodeling and wound repair.⁴ Binding of Angiotensin II, the main effector hormone, to the angiotensin II type 1 receptor (AT1R) increases the production of inflammatory molecules and regulates multiple steps in the inflammatory process.⁵ Inappropriate activation of the RAS may contribute to chronic inflammation, but how this occurs is not fully understood.

Autoantibodies that react with the AT1R (AT1RaAbs) and increase expression and function of the receptor in an apparent positive feedback loop have been identified in pre-eclampsia⁶ and have been linked to transplant rejection and malignant hypertension.^{7, 8} A pro-inflammatory cascade, initiated by AT1RaAbs, has been proposed which involves antibody binding, signal transduction through ERK1/2, an increase in reactive oxygen species, NADPH oxidase expression, and nuclear factor-kappa B (NFκB) activation.⁹ Activation of this pro-inflammatory pathway has been suggested as a potential mechanism for endothelial damage observed in persons with hypertension and cardiovascular disease.^{10, 11}

A frailty phenotype, assessed using a validated clinical tool, identifies older adults who are particularly vulnerable to a host of adverse health outcomes including falls, functional decline, disability, cardiovascular events and early mortality. Substantial evidence supports associations between this phenotype and markers of chronic inflammation (neutrophil, monocyte and total white blood cell counts and IL-6 levels).¹² How inflammatory activation is sustained remains poorly defined, but multiple chronic disease states, redox imbalance, senescent cells, and increased body fat have been suggested.^{13, 14}

We hypothesized that levels of AT1RaAb would be associated with increased inflammation, deficits in functional measures (grip strength and walking speed) and negative outcomes (increased frailty and falls). This hypothesis was tested in a primary population of community dwelling individuals (the Baltimore group). A secondary group (the Chicago group) was employed to validate associations between AT1RaAb levels and functional measures and outcomes observed in the Baltimore group. Angiotensin receptor blockers (ARBs) provide cardiovascular protection, in part through their vascular and anti-inflammatory effects ¹⁵. Whether ARB treatment can modify the associations between AT1RaAb and functional measures and/or outcomes is unkown. The Chicago group, half of whom were ARB-treated, was also used to explore how ARB treatment impacts AT1RaAb associations.

METHODS

RESULTS

The epitope of AT1RaAbs has been localized to the sequence AFHYESQ in the extracellular loop 2 region (Figure 1A).²⁴ The binding of serum components to AFHYESQ was investigated by surface plasmon resonance where the peptide was immobilized on the chip surface and different dilutions of human serum were injected in the mobile phase. Sensorgrams exhibiting the classic association and dissociation phases of a binding interaction were seen in each plot (Figure 1B). The same serum sample was used to probe a Western blot of SDS PAGE resolved cell lysates from cell lines either known to express or not express AT1R. The major immunoreactive band exhibited a M_r of 43 kDa similar to AT1R while the negative control revealed no immunoreactive bands (Figure 1C). To quantify the levels of AT1RaAb in serum, an antigen capture ELISA was developed. The assay used a rabbit anti-FHYESQ IgG antibody as a standard and conditions which gave a reproducible linearity up to 133 ng/ml (Figure 1D). The peptide AFHYESQ was able to block the binding of serum autoantibodies to microtiter wells, yielding a dose dependent reduction in levels of AT1RaAb measured by the ELISA in competition assays (Figure 1E).

Demographic and clinical variables of the discovery and validation groups are presented in Table 1. For both groups, individuals over 70 years of age were more likely to be female and had increased values for systolic blood pressure, number of comorbidities, BMI and IL-6. For the Baltimore group, Serum levels of AT1RaAb exhibited a median value of 4.47 (range 0.31 – 20.3) µg/ml and a varied scatter when analyzed as a function of age (Figure 2 A). When AT1RaAb values were grouped by decade of life, there was no significant change in median values until after age 70 (Figure 2 B). A Kruskal-Wallis ANOVA with Dunn’s multiple comparisons test assessed that AT1RaAb levels from subjects in their 70s and 80s were significantly higher then those in earlier decades, P < 0.0001. To further investigate this jump in AT1RaAb levels in adults over the age of 70, this group (“older adult”) was further segregated using the Fried frailty criteria to distinguish between low risk and high risk individuals. High risk older adults had higher median levels, 9.0 (range 2.8 – 20.3) µg/ml, of serum AT1RaAb as compared to either the adult or low risk older adults, 3.8 (range 0.3 – 18.5) µg/ml and 6.0 (range 1.6 – 17) µg/ml, respectively (Figure 2 C).

ROC curve analysis for detecting high risk versus low risk status in older adults yielded an area under the curve of 0.76 ± 0.05 and, at a cut-off value of 8.2 µg/ml, a specificity of 80% (95% CI 69 – 88) and sensitivity of 61% (95% CI 41 – 78). Based on the ROC curve analysis, on average the high risk (frail) individuals will be correctly classified as such 76% of the time using the 8.2 µg/ml AT1RaAb cutoff. Logistic regression analysis revealed that individuals with AT1RaAb levels above 8.2 µg/ml were 3.9 (95% CI 1.38 – 11.0) times more likely to be at high risk after adjusting for age (P < 0.05). When analyzed in pre-specified age groups, individuals aged 70 – 75, with AT1RaAb ≥ 8.2 were 4.88 (95% CI 1.50 – 18.86) times more likely to be at high risk (P < 0.01) compared with those with AT1RaAb < 8.2 µg/ml. Among individuals who are older than 75 years of age, those with AT1RaAb ≥ 8.2 µg/ml were 9.0 (95% CI 1.3 – 63.8) times more likely to be at high risk compared with those with AT1RaAb < 8.2 µg/ml (P < 0.05).

To further characterize the differences between the adult and low and high risk older adult groups, the distribution of IL-6, BMI and blood pressure were analyzed (Figure 2 D–G). Individuals over the age of 70 exhibited significantly elevated IL-6, BMI and systolic BP compared with adults below the age of 70. High risk individuals had higher levels of serum IL-6 with a median value of 1.7 (range 0.7 – 9.0) pg/ml versus 1.1 (range 0.5 – 3.2) pg/ml for low risk subjects, P < 0.01. High risk older adults also had significantly higher median BMI values when compared to low risk older adults. The association between AT1RaAb and IL-6, BMI and BP was investigated (Figure 3). Across all three groups (adults and older adults at low or high risk) there was a strong positive correlation between AT1RaAb and IL-6 (Spearman r = 0.33, P < 0.0001). Serum levels of AT1RaAb were significantly associated with higher BMI (Spearman r = 0.36, P < 0.0001) and systolic BP (Spearman r = 0.28, P < 0.0001). In the Baltimore group, the serum levels of AT1RaAb were also statistically significantly correlated with a history of heart disease (Spearman r = 0.26, P < 0.0005).

Older adult high risk status has been strongly associated with weaker grip strength and decline in walking speed as well as increased falls and mortality.^{16, 25} The association between AT1RaAb levels and functional measures and outcomes was investigated in low and high risk older adults from the primary population in Baltimore (Figure 4 A – C). The level of AT1RaAbs correlated negatively with grip strength (Spearman r = −0.34, P < 0.01), walking speed (Spearman r = −0.30, P < 0.05), and positively with number of falls in older adults (Spearman r = 0.40, P < 0.005). The mean grip strength was 30 ± 12 kg for subjects with < 8.2 µg/ml AT1RaAb levels and 19 ± 7.7 kg for subjects with higher AT1RaAb levels (P = 0.0005 by unpaired t test with Welch’s correction). The mean walking speed was 1.07 ± 0.40 m/s for subjects with < 8.2 µg/ml AT1RaAb levels and 0.76 ± 0.33 m/s for subjects with ≥ 8.2 µg/ml AT1RaAb levels (P < 0.005 by unpaired t test with Welch’s correction).

The number of subjects who had 1 or more falls was 12 and 16 , respectively, for subjects with < 8.2 µg/ml or ≥ 8.2 µg/ml AT1RaAb. The mean number of falls per subject was 0.80 ± 1.1 for subjects with < 8.2 µg/ml AT1RaAb levels and 1.70 ± 1.80 for subjects with ≥ 8.2 µg/ml AT1RaAb levels (P < 0.05 by unpaired t test with Welch’s correction). Logistic regression of falls as a function of AT1RaAb levels yielded an odds ratio of 1.25 (95% CI 1.06 to 1.47) unadjusted, 1.28 ( 95% CI 1.07 to 1.47) for age adjusted and 1.33 (95% CI 1.06 to 1.66) for age, gender, BMI and BP adjusted models. For every 1 µg/ml change in AT1RaAb levels, the odds of falling increased approximately 30%. Mortality was too low (6%) in the Baltimore group to enable modeling of the association between time to death and serum AT1RaAb levels.

For the Chicago group, samples from 60 subjects, half treated with ARBs, had their associated demographic and physiologic covariates, physical function tests, health outcomes, and AT1RaAb levels analyzed. The Chicago group had a higher average age, proportion of females, number of comorbidities, and reduced grip strength and walking speed compared with the Baltimore group (Table 1). Treatment with ARBs was associated with better control of systolic blood pressure in patients with high AT1RaAb as compared to patients with low AT1RaAb (128 ± 21 versus 148 ± 17 mm Hg, P < 0.05). Association similar to those seen in the Baltimore group were observed in the Chicago group for grip strength (Spearman r = −0.57, P < 0.005), walking speed (Spearman r = −0.47, P< 0.005) and falls (Spearman r = 0.30, P < 0.05) (Figure 4 D – F). Comparing functional measures by the AT1RaAb cutoff (< 8.2 versus ≥ 8.2 µg/ml) yielded 18 ± 4.5 versus 12 ± 5.8 kg for grip strength (P < 0.005 by unpaired t test with Welch’s correction) and 0.66 ± 0.22 versus 0.48 ± 0.14 m/s for walking speed (P < 0.05 by unpaired t test with Welch’s correction. The number of subjects who had 1 or more falls was 21 and 25 , respectively, for subjects with < 8.2 µg/ml or ≥ 8.2 µg/ml AT1RaAb. The mean number of falls per subject was 1.5 ± 1.2 for subjects with < 8.2 µg/ml AT1RaAb levels and 1.8 ± 1.6 for subjects with ≥ 8.2 µg/ml AT1RaAb levels (not significantly different). Logistic regression of falls in the Chicago group as a function of AT1RaAb levels yielded an odds ratio of 1.31 (95% CI 1.01 to 1.69) though any adjustments made the association no longer significant.

These observations of an association between generalized weakness (slower walking speed and weaker grip strength) and higher AT1RaAb levels led to an investigation of the impact of AT1RaAbs on outcomes in the Chicago group. Regression analysis of the natural logarithm of time to death in the Chicago group yielded an unadjusted β_AT1RaAb of −0.097 (95% CI −0.150 to −0.044) with P < 0.005. Adjusting for age, gender and BMI did not significantly alter the association as β_AT1RaAb was −0.096 (95% CI −0.255 to −0.036) with P < 0.01. For every 1 µg/ml increase in AT1RaAbs, the time to death decreased by 9% after adjusting for age, gender, BMI and BP.

The associations of AT1RaAbs with functional and outcome measures in older individuals may by modified by ARB treatment since ARBs may compete with AT1RaAbs for receptor occupancy. Multiple-adjusted models were used to explore the effect of ARB treatment on the association between AT1RaAbs and functional measures and outcomes. Among older adults not receiving ARBs, there was an average 2.6 (95% CI 0.52 – 4.68) kg decrease in grip strength for every 1 µg/ml increase in AT1RaAb level after adjusting for age. Multiple adjustments including gender, comorbidities, IL-6 and BMI, in addition to age did not affect the association in older adults who did not receive ARBs - a decrease of 2.8 kg (95% CI 0.67 – 4.93) for every 1µg/ml increase in AT1RaAb, P <0.01. This relationship was not significant in older individuals receiving ARB treatment after age or other adjustments. The data is consistent with ARB treatment attenuating the negative effects of AT1RaAb on grip strength. The association between AT1RaAb levels and the natural logarithm of time to death was analyzed by multiple regression and split by ARB treatment and adjusted for age, gender, BMI and BP. A β_AT1RaAb of −0.097 (95% CI −0.0158 to −0.036) and P < 0.0001 was assessed for subjects not receiving ARBS, while a β_AT1RaAb of 0.024 (95% CI −0.057 to 0.106) and P = 0.53 was obtained for those who were treated with ARBs. Consistent with the grip strength data, ARB treatment attenuated the association between time to death and AT1RaAb levels. These observations were unlikely to be caused by demographic or clinical differences between those on ARBs and not on ARBs as the characteristics of Chicago group segregated by ARBs treatment, with the exception of thyroid disease, were similar (Table 2).

Elevated serum levels of AT1RaAbs have been clinically associated with pre-eclampsia.^{9, 23, 24} The potential impact of past pregnancies and history of preeclampsia as factors that could alter AT1RaAb levels in our two groups (Chicago and Baltimore) was investigated. Participants enrolled in the Chicago group were recruited as part of the Chicago Religious Order Study. Women enrolled in this study were nuns with no reported history of pregnancy and/or children. Participants from the Baltimore population were surveyed for pregnancy history. There was no reported pre-eclampsia in female Baltimore participants, which was not surprising, given that the prevalence of pre-eclampsia is around 3% of all pregnancies in the U.S.²⁶ Furthermore, adjusting for past pregnancy history did not alter associations between AT1RaAb levels and functional measures. This was also supported by the observation that adjusting for gender did not change the association of AT1RaAb with the outcomes reported.

DISCUSSION

In two distinct groups higher serum AT1RaAb levels were strongly associated with worse blood pressure control, worse functional measures and higher rates of falls and death.

To date the relationship between RAS activation and lifespan remains unclear.²⁷ RAS blockade in some animal studies results in a 25% extension of lifespan ²⁸ as well as improved physical performance.²⁹ However, the response to RAS blockade has varied widely in human studies of variable design.³⁰ Our data suggest thatARB treatment may modulate associations between AT1RaAb levels and physical performance measures and mortality. If these associations are is confirmed, studies of RAS blockers might target individuals with high levels of these autoantibodies.

The correlation between IL-6, systolic BP, and serum AT1RaAb is in agreement with previous research documenting the role of inflammation in the development of hypertension and with a role for AT1RaAb as a pro-hypertensive autoantibody.^{31, 32} It is not clear whether AT1RaAbs are specific disease-causing agents, or if their levels reflect age-related global dysregulation of autoantibody production. Autoantibodies in general are capable of stimulating receptors and causing disease, such as Graves’ disease ³³ and dilated cardiomyopathy.³⁴ In our study we observed an association between serum level of AT1RaAb and: the pro-inflammatory cytokine IL-6 which has been linked to frailty; ^{35, 36} functional measures associated with risk status (walking speed and grip strength); and poor outcomes (falls, death). The correlation coefficients for these associations were significant, though the percent of variance in the outcome measure accounted for might be considered small, ranging from 9 to 32%. Grip strength, walking speed and falls are complex traits and correlations in the range of what we report are considered biologically significant.^{37, 38} For example, the In Chianti study of inflammatory biomarkers and physical performance in 1020 participants aged 65 years and older found that CRP, IL-6, and IL1RA were significantly correlated with physical performance (Spearman r = −0.162, r = −0.251, and r = −0.127, respectively) and walking speed (Spearman r = −0.143, r = −0.254, r = −0.133, respectively). A similar correlation (r = 0.20) was seen in a study of myostatin and grip strength and a biological linkage between myostatin and the outcome is evident.³⁹ Differences in serum levels of these autoantibodies may explain some of the variability in results of studies of RAS blockade for improving muscle strength and physical performance in older individuals.⁷

Our exploratory data indicate that the use of ARBs attenuated the association between levels of AT1RaAb and decline in grip strength and time to death. Hints of this effect may be gleaned from previous reports documenting that ARBs may retard the decline in muscle strength and physical activity in older hypertensive persons.⁴⁰ Our data suggests that this effect may have been partially explained by the presence of AT1RaAb in these older adults. Analysis of the effect of ARB treatment on the association between autoantibody levels and number of falls yielded wide confidence intervals to the odds ratio that lost significance with any adjustment. The increased incidence of falls and serious injuries with the use of ARBs has been previously reported ⁴¹ and this may have confounded our analysis. It may also be that the limited number of samples in the Chicago group negatively impacted the power of the analysis.

The present study and reports on the effects of AT1RaAb on cytokines, reactive oxygen species and endothelial dysfunction, ^9–11 extends a line of evidence suggesting that elevated serum levels of AT1RaAb may have contributed to the precipitation and progression of physical decline in those older individuals at high risk for negative health outcomes (increased falls and mortality) and may do so via specific pro-inflammatory effects. Further, patients with high levels of antibodies that stimulate AT1R may be more likely to benefit from ARB treatment that blocks AT1R. The limitations to this study include a relatively small validation group sample size that, while yielding statistical significance in our primary outcomes, impeded robust analyses of ARB effects on falls and mortality by subgroups. Perhaps more importantly, the study is designed to ascertain associations, not causality. Larger studies are required to fine tune the utilization of AT1RaAb as a screening tool and large intervention trials are needed to guide selection of individuals that might benefit most from RAS blockade.

In summary, the association of risk status with serum levels of AT1RaAb provides one plausible biological pathway for RAS-mediated chronic inflammation and subsequent disease progression that can be probed for therapeutic intervention. ARB treatment was associated with a blunting of the harm seen in patients with high levels of agonistic autoantibodies to AT1R.