Streptococcal Serology in Acute Rheumatic Fever Patients

Findings From 2 High-income, High-burden Settings

Abstract

Background: 

Globally, there is wide variation in streptococcal titer upper limits of normal (ULN) for antistreptolysin O (ASO) and anti-deoxyribonuclease B (ADB) used as an evidence of recent group A streptococcal infection to diagnose acute rheumatic fever (ARF).

Methods: 

We audited ASO and ADB titers among individuals with ARF in New Zealand (NZ) and in Australia’s Northern Territory. We summarized streptococcal titers by different ARF clinical manifestations, assessed application of locally recommended serology guidelines where NZ uses high ULN cut-offs and calculated the proportion of cases fulfilling alternative serologic diagnostic criteria.

Results: 

From January 2013 to December 2015, group A streptococcal serology results were available for 350 patients diagnosed with ARF in NZ and 182 patients in Northern Territory. Median peak streptococcal titers were similar in both settings. Among NZ cases, 267/350 (76.3%) met NZ serologic diagnostic criteria, whereas 329/350 (94.0%) met Australian criteria. By applying Australian ULN titer cut-off criteria to NZ cases, excluding chorea, ARF definite cases would increase by 17.6% representing 47 cases.

Conclusions: 

ASO and ADB values were similar in these settings. Use of high ULN cut-offs potentially undercounts definite and probable ARF diagnoses. We recommend NZ and other high-burden settings to use globally accepted, age-specific, lower serologic cut-offs to avoid misclassification of ARF.

Acute rheumatic fever (ARF), an autoimmune complication of infection with group A streptococci (GAS) and its sequel, rheumatic heart disease (RHD), are causes of major morbidity and mortality globally in low-resource settings,¹ and in Indigenous communities within Australia and New Zealand (NZ).² Cases of ARF are notifiable in NZ and in selected Australian states and territories. A persisting challenge in ARF diagnosis is the absence of a definitive diagnostic test that remains an impediment to clinical management, surveillance and research. The Jones criteria provide a clinical guide to diagnosis³ alongside laboratory evidence of preceding GAS infection, ideally by demonstrating at least a 2-fold rise in streptococcal titer between acute and convalescent sera samples taken 14–28 days apart.^4,5 In reality, sequential venepuncture may not be possible, and depending on the timing of the ARF episode and background streptococcal exposure, a rise in titer may not be demonstrable. Using upper limits of normal (ULN) cut-offs, typically the 80th percentile of titer values in a healthy population, are recommended when paired sera are unavailable.⁴

Challenges in establishing ULN are well described.⁶ These arise due to different characteristics of available assays because titers of antistreptolysin O (ASO) and anti-deoxyribonuclease B (ADB) remain elevated for many weeks after infection,⁷ and because subclinical infection means that a proportion of “healthy” populations may have elevated values. Furthermore, non–group A β-hemolytic streptococci can also cause elevated ASO titers.⁵ Titer cut-offs determined from adult populations may be inappropriately low for pediatric populations.^6,8 In settings with a high burden of ARF, diagnostic sensitivity is important to ensure that the often-subtle diagnosis of ARF is not missed.

Different ASO and ADB titer ULN recommendations are shown in Table 1. Previously it was recommended that appropriate ULN be established in each population among healthy individuals.¹⁸ However, in a 2009 Fijian study, Steer et al¹³ reported ASO and ADB ULN similar to those in Australia and the United States, concluding that a uniform age-specific ULN for streptococcal serology might be applicable globally. Consequently, Australia adopted Fiji’s ULN in their 2012 updated ARF diagnosis and management guideline.⁸ In NZ, ULN titers were established from a convenience sample of children <15 years of age hospitalized for reasons other than ARF or RHD in Auckland in 1982.⁹ The single all-age ULN for ASO and ADB titers is higher than recommended elsewhere in the world (Table 1).

TABLE 1.: Recommended Upper Limits of Normal for ASO and ADB Titers

Concern has been expressed about whether currently recommended ULN in some ARF-endemic settings may be too high, such that genuine ARF cases might be missed. Although previous studies have reported average ASO and ADB values for an ARF cohort,¹⁹ actual ASO and ADB values among individuals diagnosed with ARF have not to our knowledge been reported previously in NZ or Australia. In this study, we aimed to: (1) determine the proportion of cases fulfilling diagnostic guidelines in NZ and Australia’s Northern Territory (NT), respectively; (2) determine whether the currently recommended local guidelines for ASO and ADB are being applied and (3) calculate the proportion of cases fulfilling alternative serologic diagnostic criteria.

METHODS

NZ Audit

We undertook a retrospective audit of serologic results from all rheumatic fever cases notified with an onset date from January 2013 to December 2015. Data included age, sex, ARF onset date, clinician classification, first episode or recurrence, major and minor clinical manifestations and ASO and ADB titers. Classification of definite, probable and possible ARF was assigned using the modified Jones criteria where titer levels of ASO ≥480 IU/mL and/or ADB ≥680 U/mL indicate serologic evidence of a preceding GAS infection for all ages.¹⁰ NZ guidelines allow GAS detected on throat swabs as evidence of a preceding GAS infection, although a positive culture alone demotes a case to probable or possible ARF. Chorea, with alternative etiologies excluded, is a single criterion for definite rheumatic fever.^8,10

No titer dates were recorded, so the highest recorded titer was used. ASO was typically assayed by immunonephelometry (Beckman Coulter IMMAGE 800 Beckman Coulter, Indianapolis) or turbidimetric technique using the human antistreptolysin-O kit on a SPAplus analyser (The Binding Site, Birmingham, CA). ADB was typically assayed by immunonephelometry (Beckman Coulter IMMAGE 800) or an enzyme inhibition assay (bioMerieux, Marcy l’Etoile, France).

We applied NZ clinical case definitions with NZ ULN cut-offs, compared with Australian ULN cut-offs.

Australian NT Audit

We undertook a retrospective audit of serologic results from consecutive patients included in the NT register from January 2013 to December 2015, with a diagnosis of ARF. The same data were collected as for NZ, with the addition of serology dates. Classification of definite, probable and possible ARF was assigned using the Australia guidelines⁸ which provide criteria for high-risk populations.³ The only differences with NZ are that polyarthralgia is a major criterion in Australia and a minor criterion in NZ, and monoarthralgia is a minor criterion in Australia and not included in NZ (Table 2). ¹⁰ Note, in Australia, categories of definite, probable and possible ARF are referred to respectively as “definite,” “probable (highly suspected)” and “probable (uncertain).”⁸ Age-specific serologic ULN applicable in Australia is shown in Table 1. The peak ASO or ADB result within 1 month of the ARF diagnosis date (<30 days before <30 days after) was used. ASO was assayed by immunonephelometry and ADB using particle-enhanced immunonephelometry on a Siemens BN II Nephelometer (Siemens Healthcare GmbH, Erlangen, Germany).

TABLE 2.: Clinical Characteristics of Acute Rheumatic Fever Cases, January 2013–December 2015

Statistical Analyses

Descriptive and comparative statistics were calculated using Stata 14.0. (Stata Corp, College Station, TX). Proportions were compared using χ² 2-tailed tests. Nonparametric data were log-transformed where appropriate (age) and compared using Student t test, or compared using Wilcoxon rank sum test (serologic titers).

Ethical Approval

The Health and Disability Ethics Committees, NZ, waived ethical approval since the audit met exemption criteria. The Human Research Ethics Committee of the NT Department of Health and Menzies School of Health Research approved the Australian site as a low-risk audit (#2016–2604).

RESULTS

NZ Serology Audit Results

There were 350 notified cases with ARF onset in 2013–2015 with streptococcal serology results. These 350 cases included 323 first-episode cases and 27 recurrent cases (Table, Supplemental Digital Content 1, https://links.lww.com/INF/D298). Clinical features are shown in Table 2. Geometric mean age was 12.3 years [95% confidence interval: 11.7–12.9].

The median peak titers overall were as follows: ASO 562 IU/mL [interquartile range (IQR): 337–754] and ADB 599 U/mL (IQR: 300–900) (Fig. 1). At least 1 ASO level was available for all 350 cases and at least 1 ADB level for 338 cases. Using NZ ULN criteria, 208/350 individuals (59.4%) with ARF had elevated ASO compared with 167/350 (47.7%) with elevated ADB (P = 0.008).

FIGURE 1.: ARF cases serology results showing maximum recorded value per patient. A: New Zealand ASO titers. B: New Zealand ADB titers. C: Northern Territory ASO titers. D: Northern Territory ADB titers.

Using the Australian age-specific ULN cut-offs for NZ patients increased the proportion fulfilling serologic ARF diagnostic criteria from 267/350 (76.3%) to 329/350 (94.0%), and from 243/308 (78.9%) to 297/308 (96.4%) when excluding chorea cases (Fig. 2A; Table, Supplemental Digital Content 1, https://links.lww.com/INF/D298). By applying Australian ULN titer cut-off criteria to NZ cases, excluding chorea, ARF definite cases would increase by 17.6% representing 47 cases over 3 years.

FIGURE 2.: Acute rheumatic fever cases grouped according to different diagnostic criteria. The bars show percentages, with the number of cases shown in text beside the bars. New Zealand cases (A) and Australian cases (B) with different diagnostic criteria applied. Cases with chorea were removed.

Twenty-four individuals not meeting serologic criteria fulfilled positive throat culture criteria for probable/possible ARF giving 291/350 (83.1%) cases fulfilling serologic and/or positive throat culture criteria. Furthermore, 15 cases had chorea without serologic/culture evidence of GAS, giving 306/350 (87.4%) fulfilling acceptable NZ serologic, throat culture or chorea clinical diagnostic criteria.

Serial results were available in 162 individuals for ASO and 159 for ADB titers. Just over half had a rise or no change in titer for ASO (54.3%), and ADB (54.1%), with the remainder showing a reduction in titer. Six of 162 (3.7%) paired ASO and 11/159 (6.9%) paired ADB tests demonstrated a 2-fold rise (Fig. 3A). All cases with a 2-fold ASO rise also reached the NZ ULN. Seven cases with a 2-fold ADB rise did not reach the NZ ULN although 5 had met the ASO ULN.

FIGURE 3.: ASO and ADB titer levels for ARF cases. A: Serial ASO and ADB titers for New Zealand ARF cases. B: Serial ASO titer levels for Northern Territory ARF cases. Serial ASO and ADB titers in the New Zealand cohort represented as a heat map. The ULN titer is shown in white with titers below the ULN in blue and titers above the ULN in red. Paired samples in which there was a ≥2-fold rise are indicated by the square bracket.

Australian NT Results

Data for 196 patients were accessible. One hundred eighty-two had a streptococcal serology result within the appropriate time frame including 130 first-episode cases and 52 recurrent cases (Table, Supplemental Digital Content 2, https://links.lww.com/INF/D299). Clinical features are shown in Table 2. Geometric mean age was 12.3 years (95% confidence interval: 11.4–13.3), similar to the NZ cohort (P = 0.97).

The median peak titers were as follows: ASO 610 IU/mL (IQR: 400–913) and ADB 600 U/mL (IQR: 400–850) (Fig. 1); very similar to NZ. ASO was more commonly elevated than ADB. Using Australian age-specific ULN, 171/182 individuals (94.0%) had elevated ASO compared with 119/182 (65.4%) with elevated ADB (P = 0.042). Overall, 175/182 cases (96.2%) fulfilled Australian serologic criteria, whereas 129/182 (70.9%) fulfilled NZ criteria (Table, Supplemental Digital Content 2, https://links.lww.com/INF/D299).

Among cases of definite ARF with nonchorea manifestations, 116/119 (97.5%) met Australian criteria. Of those, 78/119 (65.5%) had elevated titers of both ASO and ADB; 36 (30.3%) had elevated ASO alone, and 2 (1.7%) had elevated ADB alone. The proportion of nonchorea definite ARF cases meeting NZ serologic criteria was 79/119 (66.4%) (Fig. 2B and Table, Supplemental Digital Content 2, https://links.lww.com/INF/D299).

Contrasting with NZ findings, ADB was significantly lower (median: 498; IQR: 280–740) among recurrences than first episodes (median: 600; IQR: 420–949), P = 0.003. However, there was no difference in ASO: 641 (IQR: 360–818) in recurrences versus 600 (IQR: 400–913) in first episodes (P = 0.693).

Serial results were available in 27 individuals with a median interval of 5 days (range: 0–42 days). Most had a small rise or no change in streptococcal titers. Only 2/27 (7%) had a 2-fold rise in titer (Fig. 3B).

DISCUSSION

In this audit of GAS serology results among patients diagnosed with ARF in diverse settings, we found that strict application of the NZ serologic diagnostic criteria would result in undercounting of definite ARF. However, clinicians in NZ appear to be diagnosing ARF appropriately, despite patients’ serologic results not meeting the high NZ ULN thresholds, with clinical concordance with diagnostic and serologic guidelines at 87.4% in NZ cases, compared with 96.2% in Australia’s NT cases. Because we only audited patients who were notified or registered, we are unable to determine whether ARF cases are genuinely being missed, leading to lost opportunities for secondary prevention of ARF to prevent progression to RHD.

If NZ were to lower the streptococcal titer ULN cut-offs for the diagnosis of ARF, the impact of overdiagnosis would need to be considered. However, a recent audit (2010–2015) of RHD hospitalizations in NZ identified 20–40 cases per year of RHD in those <40 years of age without previous ARF diagnosis or hospitalization.²⁰ This surprisingly high number suggests a need to improve ARF diagnosis through improved recognition in primary care and by reviewing laboratory criteria.

By applying the Australian ULN titer cut-off criteria to NZ cases excluding chorea, ARF definite cases would increase by 17.6% representing 47 cases over 3 years. In NZ, those patients in whom a diagnosis of probable or possible ARF is made are still followed and prescribed secondary prophylaxis with penicillin, albeit for a shorter duration for possible cases. NZ has an effective system for enrolling ARF cases into treatment programs and ensuring secondary prophylaxis is adhered to, resulting in relatively few ARF recurrences in the 3-year period reviewed (27 instances of ARF recurrence, 7.7% of all ARF cases) compared with NT where adherence is improving but remains suboptimal with 52 instances of ARF recurrence, 28.6% of all ARF cases.

We found that peak ASO and ADB were not substantially different between NZ and Australia cases. The proportion of cases fulfilling diagnostic criteria at the different serologic cut-offs in NZ and the NT were similar: the NZ serologic guidelines were fulfilled by 267/350 (76.3%) of NZ cases and 129/182 (70.9%) of NT cases, whereas Australian guidelines were fulfilled by 329/350 (94.0%) of NZ cases and 175/182 (96.2%) of NT cases. These findings suggest that in the ethnically diverse populations of both settings at risk for ARF, similar ASO and ADB profiles are observed, and the same recommended ULN could be applicable across these settings.

In both the NZ and NT datasets, ARF patients were more likely to have elevated ASO than ADB. Historically, ASO was understood to be elevated after GAS pharyngeal infection, and ADB after skin infection.²¹ A previous NT study found that ADB was raised in a majority of ARF patients but ASO was frequently normal.²² This was interpreted as supportive evidence that streptococcal skin infections may have been occurring before ARF developing—consistent with other research strongly suggestive of a role for streptococcal skin infections in triggering ARF.^23–25 However, a more recent longitudinal study of 160 children over a 2-year period showed that bona fide GAS pharyngitis can induce an elevated ADB in the absence of ASO and vice versa.⁷ This suggests that the nature of the serologic response (ASO vs. ADB) may be more related to organism type and host response than site of infection.

This audit provides helpful international comparisons of the manifestations of ARF. We found differences in clinical symptoms at presentation, with a higher proportion of NT cases presenting with arthritis or arthralgia alone, and skin or subcutaneous manifestations being seen only in NZ cases. Many NT Aboriginal people have more deeply pigmented skin than the NZ ARF population (Māori and Pacific peoples), which may impede detection of erythema marginatum.²⁶ The higher proportion with arthritis and arthralgia in NT could be because polyarthralgia (painful joints without evident effusion) is a major diagnostic criterion in Australia, or could be genuine differences in clinical manifestations of these conditions in NZ Māori and Pacific peoples compared with Australian Aboriginal peoples.

Limitations of the study include being unable to access all community laboratory results, hence bacterial culture or serologic tests undertaken before or after hospitalization were not captured. NZ titer dates were unavailable so the serial interval could not be determined. In addition, some laboratories did not give precise titer levels, making interpretation of 2-fold rise in titer difficult. However, most individuals with paired serology did not demonstrate rising titers, supporting the need for robust ULN cut-offs for diagnosis of ARF in these settings.

Age and serologic characteristics of ARF patients in NZ and Australia NT appear to be similar. To align with current evidence and avoid potential undercounting of definite and probable cases of ARF, NZ should consider either updating their guidelines using the age-specific titer ULN used in Australia¹³ or conducting a study to ascertain contemporary NZ age-specific ASO and ADB titer ULN.