HCV reflex testing: A single-sample, low-contamination method that improves the diagnostic efficiency of HCV testing among patients in Alberta, Canada

Abstract

BACKGROUND

Hepatitis C virus (HCV) can be cured with antiviral treatments. Diagnosis normally requires two blood samples, one for serology screening and one for molecular confirmation. This multi-step process creates barriers in patient care and decreases testing for hard-to-reach populations. We used the cobas® 6800 to detect HCV RNA after antibody testing to investigate whether a single-sample reflex testing method is effective and efficient for diagnosing HCV-positive patients.

METHODS

HCV RNA–positive clinical samples (n = 152) were interchangeably loaded on the ARCHITECT i2000SR with negative samples (n = 152) in a checkerboard fashion, tested for HCV antibodies using fixed probes, and directly transferred to the cobas 6800 for molecular testing. Contamination rates, sensitivity, and specificity were determined by comparing Abbott m2000 and cobas 6800 viral loads. After implementing reflex testing, clinical data over a 6-month period were analyzed for diagnostic efficiency.

RESULTS

Contamination was present in 5 of 152 pairs (3.29%) after reflex testing. Sensitivity and specificity were 99.3% (95% CI 95.1% to 99.9%) and 100% (95% CI 97.5% to 100%), respectively, using the cobas 6800 assay after serotesting. Approximately 97% of clinical patients received a conclusive test result with the reflex-testing algorithm. For HCV-positive patients, mean diagnostic turnaround times were significantly lower using reflex testing versus the two-sample method (4 versus 39 days; p < 0.0001).

CONCLUSIONS

HCV reflex testing demonstrated low levels of contamination without compromising the integrity of the molecular assay. Implementation in clinical laboratories would increase the efficiency of diagnosis and decrease steps in the continuum of care for patients.

Introduction

Hepatitis C virus (HCV) is a blood-borne pathogen that preferentially replicates in the liver (1). According to the World Health Organization (WHO), more than 71 million people worldwide are infected with HCV (2). Infections are spontaneously cleared in 25%–30% of individuals, whereas 70%–75% of infections become chronic and can cause liver cirrhosis or hepatocellular carcinoma (3). Early diagnosis of an HCV infection is therefore optimal to control and treat the virus before liver damage can occur (4).

Upon infection with HCV, the immune system generates antibodies against the virus (5). Serological testing identifies previous exposure to the virus and is regularly used to screen patients for infection (6). However, antibody testing is unable to differentiate between exposure and active infection, and those with spontaneous clearance or active, chronic, or resolved infection will screen positive for HCV antibodies. Subsequently, active infections need to be identified via molecular detection of HCV RNA (7). Because of the higher sensitivity of molecular assays and the possibility of upstream contamination on the serological instrument, diagnosis typically requires two blood samples to be collected (one for serology and one for molecular testing; the two-sample method) (8).

Multiple studies have shown that patients are more likely to be cured of the virus if HCV RNA molecular testing is performed within 6 months of serology (9–11). Unfortunately, the requirement for patients to submit a secondary blood sample for HCV RNA testing means that patients are unable to receive a conclusive diagnosis at the same time as their serology result and are often lost to follow-up between tests. One method to ensure molecular testing among seropositive patients, and decrease those lost to follow-up, is to implement HCV reflex testing, in which a single blood sample is used for antibody testing and molecular confirmation if needed. However, molecular reflex testing from the same tube used for serology is often discouraged for the clinical diagnosis of HCV because of contamination concerns, particularly when fixed probes are used in upstream serology testing. In addition, there is concern over the sensitivity and specificity of molecular assays after antibody testing on the same sample (12).

In this study, we emulated the molecular reflex testing workflow on clinical samples by performing HCV antibody testing on the ARCHITECT i2000SR (Abbott Laboratories, Abbott Park, Illinois) and then transferring samples to the cobas^® 6800 (Roche Diagnostics, Rotkruez, Switzerland) for HCV RNA nucleic acid testing (NAT). We analyzed rates of contamination and calculated sensitivity and specificity to examine the feasibility of HCV reflex testing in clinical laboratories. Moreover, we assessed reflex testing results and turnaround times in the public health laboratory system to determine whether reflex testing implementation can improve HCV diagnostic efficiencies for patients in Alberta.

Materials and Methods

HCV testing in Alberta

All HCV testing in Alberta is performed at the provincial public health laboratory (Alberta Precision Laboratories, or ProvLab), which has two centralized locations: Calgary and Edmonton. HCV antibody testing is carried out at both locations, whereas HCV RNA confirmation testing is performed at Edmonton. ProvLab uses a continuous computational language (CCL) look-back method to identify patients with a history of HCV antibody or RNA positivity. Before December 12, 2019, all HCV antibody requests were performed, but confirmation of previous positives was not repeated. After December 12, 2019, all antibody requests for patients who have previously been confirmed positive for HCV are automatically cancelled, and physicians are guided to monitor patients via molecular testing. Antibody testing is performed using the Abbott Architect i2000SR and dedicated molecular RNA testing is performed using the Abbott m2000. Test results and patient demographics are uploaded to the ProvLab Laboratory Information System (LIS), and samples are kept for at least 2 years before discarding.

Sample selection

Clinical samples sent to ProvLab for routine HCV testing between December 5, 2017, and December 5, 2019, were used in this study. Selected samples were dedicated molecular serum samples and were not previously run on any other instruments besides the Abbott m2000. HCV-seropositive samples (n = 170) with low-range (<1,000 IU/mL; n = 18), mid-range (between 1,000 and 1,000,000 IU/mL; n = 12), and high-range (>1,000,000 IU/mL; n = 140) viral loads were selected for use and represented a cross-section of HCV genotypes (1a, 1b, 2, 3, and 4; not typed; or mixed typed). The mean values of HCV RNA viral loads across the low-range, mid-range, and high-range groups were determined from original viral load values (performed on the Abbott m2000). HCV-seropositive samples negative for HCV RNA (n = 147) were selected as negative controls.

HCV antibody testing

To evaluate possible contamination from the serology testing instrument, mid- and high-range viral load samples were run in pairs (n = 152) with negative samples on the ARCHITECT i2000SR. All samples were initially stored at –70°C and thawed at room temperature (20°C) before use. Greater than or equal to 1.5 mL each of known HCV RNA–positive samples followed by known HCV RNA–negative samples were loaded in a checkerboard pattern and placed on the ARCHITECT i2000SR. HCV antibody testing was carried out on the ARCHITECT i2000SR using fixed probes according to manufacturer instructions (13). Ten to 15 pairs of HCV RNA–positive and RNA–negative samples were run at a time until all samples were processed.

Comparing the cobas 6800 reflex testing with the Abbott m2000 HCV RNA output

Once HCV antibody testing of samples was completed, they were placed onto the cobas 6800 system for HCV RNA detection (14). Low-range viral load samples (n = 13; the 5 contaminated samples were excluded) were also run on the ARCHITECT i2000SR followed by the cobas 6800 to emulate the clinical workflow and to negate analyte degradation resulting from increased freeze–thaw cycles. Although the cobas 6800 platform has a diagnostic claim, off-label technology access was used to extract the quantitative viral load run in the background of every qualitative test from the instrument software. Quantitative results from the cobas 6800 were then compared with those originally recorded from the Abbott m2000 to validate our approach against a reference standard. Pairs were classified as either concordant (Abbott m2000 detected–cobas 6800 detected or Abbott m2000 not detected–cobas 6800 not detected) or discordant (Abbott m2000 detected–cobas 6800 not detected or Abbott m2000 not detected–cobas 6800 detected). Abbott m2000 detected–cobas 6800 not detected samples were assumed to be discordant as a result of viral RNA degradation. Carry-over contamination was defined as Abbott m2000 not detected–cobas 6800 detected. The sensitivity and specificity of the cobas 6800 assay was assessed after upstream serology testing. Finally, the reflex testing workflow was validated using inter- and intra-assay method validation, as previously described (15). Positive and negative pooled samples were reflex tested in triplicates for 3 days and as singles for 7 days, and a percent coefficient of variation (%CV) threshold of less than 20% was used for validation.

HCV reflex testing of clinical samples

HCV reflex testing was implemented in ProvLab on December 12, 2019. HCV antibody–positive samples with RNA viral loads of more than 1,000 IU/mL were defined as detected for HCV infection; those with undetected viral loads, as negative; and those with viral loads between 1 and 1,000 IU/mL were defined as indeterminate. The cut-off for positive detection was chosen as 1,000 IU/mL because it was 2 standard deviations above observed RNA contamination during assay validation. HCV testing data were extracted from the ProvLab LIS for 6 months before (June 12, 2019–December 11, 2019) and 6 months after (December 12, 2019–June 12, 2020) the implementation of reflex testing. Data variables collected included test results, patient identifiers, demographic information, specimen collection, submission, and result verification times. Only presumed first-time HCV antibody–positive patients were included in the analyses, which were identified by a CCL look-back using Alberta personal health numbers (PHNs) in the ProvLab software or manually in the LIS using PHN (regardless of province), name, and date of birth. Mean diagnostic turnaround times from specimen collection for HCV antibody testing to completion of HCV RNA molecular testing were calculated over the two time periods. Follow-up HCV RNA testing for the two-sample method was linked to initial HCV antibody testing using overlapping PHNs, name, and date of birth. Turnaround times were recorded in days.

Data analysis

All data were collated and graphed in Stata version 15.1 (StataCorp, College Station, Texas), and flowcharts were created in Microsoft PowerPoint 2010 (Microsoft Corporation, Redmond, Washington). Paired t-tests and non-parametric sign tests were used to compare mean viral loads between the Abbott m2000 and the cobas 6800 HCV RNA NATs. A two-sample t-test with unequal variances was used to compare mean turnaround times before and after the implementation of reflex testing (two-sample versus reflex testing method). The threshold for a significant p-value was set at α = 0.05, and 95% confidence intervals (CIs) were constructed. Statistical analyses were performed using Stata version 15.1.

Results

A total of 152 pairs of serum samples were reflex tested from the ARCHITECT serology instrument to the cobas 6800 platform (where a pair constituted two HCV antibody–positive samples: one detected and one not detected by HCV viral load testing). Of these, 5 previously not-detected samples (Abbott m2000) had detectable viral loads on the cobas 6800 (Table 1; contamination event due to carry over). Four of the five samples had detectable viral loads below the cobas 6800 limit of quantification (LOQ; titre minimum <15 IU/mL); only 1 of the entire 152 samples (0.66%) that were initially classified as not detected had a viral load above the LOQ after reflex testing. In total, 147 of 152 pairs (96.7%) had no detectable carry over contamination.

Table 1:

Abbott m2000 HCV RNA viral titres compared with cobas® 6800 viral titres after HCV antibody testing on the Abbott ARCHITECT to determine reflex testing contamination rates

Pairs*	m2000 initial viral load†	cobas 6800 viral load‡	Result	Clinical interpretation
1
i	6.05 × 106	1.02 × 107	Carry over due to serology testing	Not confirmed as a new HCV infection
ii	Not detected	2.07 × 101
2
i	3.06 × 106	3.74 × 106	Possible carry over <15 IU/mL	Not confirmed as a new HCV infection
ii	Not detected	< Titre min§
3
i	4.70 × 106	3.55 × 106	Possible carry over <15 IU/mL	Not confirmed as a new HCV infection
ii	Not detected	< Titre min§
4
i	1.87 × 106	2.32 × 106	Possible carry over <15 IU/mL	Not confirmed as a new HCV infection
ii	Not detected	< Titre min§
5
i	4.66 × 106	2.72 × 106	Possible carry over <15 IU/mL	Not confirmed as a new HCV infection
ii	Not detected	< Titre min§
6–152, mean (SE), p < 0.0005¶, **	4.37 × 106 (3.49 × 105)	5.14 × 106 (3.80 × 105)	No carry over detected	Results consistent with m2000

*Pairs 1–5 are those with observed carry over and pairs 6–152 are those with no carry over detected

† Viral loads are depicted for the output from the i) positive and ii) negative samples initially run on the Abbott m2000

‡Viral loads are depicted for the output from the cobas 6800 after being reflexed from the Abbott ARCHITECT

§15 IU/mL)

¶Mean viral loads for positive samples without detectable carry over from reflex testing and standard error

**p-value compares mean viral loads between the Abbott m2000 and cobas 6800 (statistical analysis performed using a paired t-test.) p < 0.05 is significant

HCV = Hepatitis C virus; Titre min = Titre minimum

To ensure that the sensitivity of viral load testing was maintained during upstream serology testing, viral loads from 165 of the 170 seropositive samples, including the low-range viral load samples and excluding the 5 samples with detected carry-over contamination, were compared between the Abbott m2000 and the cobas 6800. Overall, there was a difference in mean viral titres between the two testing platforms, with original values on the Abbott reported to be significantly lower than those detected on the cobas (Table 1; mean 4.37 × 10⁶ IU/mL, versus mean 5.14 × 10⁶ IU/mL, p < 0.0005, respectively). In 104 of 165 samples (63.0%), the recorded viral load was higher on the cobas 6800 than on the Abbott m2000. This was particularly seen with high-range viral load samples, which were significantly higher on the cobas 6800 (high range: Abbott m2000 mean 4.55 × 10⁶ IU/mL, cobas 6800 mean 5.26 × 10⁶ IU/mL, p = 0.0001; Figure 1). However, there was no significant difference in mean viral loads between low-range or mid-range samples when tested on the cobas 6800 compared with the Abbott m2000 (low-range: Abbott m2000 mean 2.36 × 10² IU/mL, cobas 6800 mean 2.94 × 10² IU/mL, p = 0.27; mid-range: Abbott m2000 mean 5.06 × 10⁵ IU/mL, cobas 6800 mean 5.46 × 10⁵ IU/mL, p = 0.07; Figure 1).

Figure 1:

Comparison of low-, mid-, and high-range HCV viral load outputs from paired samples on the cobas® 6800 and Abbott m2000 systems*

*p < 0.05 is significant

HCV = Hepatitis C virus; X̄ = mean

A total of 288 samples (141 low-, mid-, and high-range RNA-positive and 147 RNA-negative samples) were used to assess test characteristics of the cobas 6800 when used in a reflex testing workflow (Table 2). When comparing results, 287 of 288 samples were concordant between the Abbott m2000 and the cobas 6800 (agreement 99.7%; 95% CI 98.1% to 100%). The one discordant sample had a viral load of 61 IU/mL on the Abbott m2000 but was not detected with the cobas 6800 (sensitivity 99.3%, 95% CI 95.1% to 99.9%). The negative predictive value was 99.3% (95% CI 95.4% to 99.9%). All HCV RNA–negative samples were concordant between the two assays (specificity 100%, 95% CI 97.5% to 100.0%) with a positive predictive value of 100% (95% CI 97.5% to 100.0%). To validate the laboratory workflow in our study, we measured the repeatability (intra-assay) and reproducibility (inter-assay) of the reflex testing method across two different testing sites (Table 3). The %CV for repeatability and reproducibility was below the accepted coefficient of variation threshold at 4.80% and 0.57%, respectively, validating our reflex testing workflow for use in the clinical laboratory.

Table 2:

Test characteristics of the cobas® 6800 system when used within the HCV reflex testing workflow

Outcome or characteristic	Result
Outcome, no.*
True positive	140
False negative	1†
False positive	0
True negative	147
Test characteristic, % (95% CI)‡
Sensitivity	99.3 (95.1 to 99.9)
Specificity	100 (97.5 to 100.0)
PPV	100 (97.5 to 100.0)
NPV	99.3 (95.4 to 99.9)
Agreement	99.7 (98.1 to 100)

*Results represent the number of samples from the cobas 6800 assay after reflex testing, compared with the Abbott m2000 gold standard reference method, and categorized by outcome

† Abbott m2000 = 61 IU/mL; cobas 6800 = not detected

‡95% CIs are reported with upper and lower interval values

HCV = Hepatitis C virus; PPV = Positive predictive value;

NPV = Negative predictive value

Table 3:

Intra-assay repeatability and inter-assay reproducibility analyses of the reflex testing workflow in Alberta, Canada

Assay	Sample size, n	Method	Viral load (Log10), mean (SD [σ])	%CV
Intra-assay (repeatability)	20	Run in triplicate on 3 days and in single on 7 additional days	4.59 (0.22)	4.80
Inter-assay (reproducibility)	20	Run in triplicate on 3 days and in single on 7 additional days, across two sites	4.44 (0.014)	0.57

%CV = Percent coefficient of variation

To evaluate the performance of HCV reflex testing in our patient population, testing data were analyzed 6 months after implementation of the reflex testing method in ProvLab. In total, 105,845 specimens were tested for HCV antibodies (anti-HCV); 1,478 specimens screened positive, and 1,472 of those (99.6%) were of sufficient volume for molecular reflex testing, corresponding to 1,414 unique patients presumed to be first-time HCV positives on the basis of CCL look-back (Figure 2A). Of the 1,472 specimens reflex tested, 890 (60.5%) were RNA negative, 536 (36.4%) were RNA positive (viral load >1,000 IU/ml), and 46 (3.13%) were indeterminate (viral load ≤1,000 IU/mL). Of the 46 indeterminate specimens, 3 were patient duplicates and were removed from further analysis, leaving 43 unique indeterminate patients with a mean recorded viral load of 351.1 IU/mL (SD 43.7 IU/mL, range 26–973 IU/mL; Figure 2B). For patients with an indeterminate result, 30 of 43 (69.8%) submitted a dedicated second blood sample and received follow-up HCV quantitative RNA testing. Of those, 26 were HCV RNA positive (mean viral load from the cobas 6800 400.9 IU/mL, SD 60.9 IU/mL, range 30–973 IU/mL) and 4 were HCV RNA negative (mean viral load from the cobas 6800 79.3 IU/mL, SD 30.9 IU/mL, range 26–152 IU/mL). It is possible that the 4 patients with negative follow-up testing could have initially screened indeterminate because of contamination from upstream serology testing or could represent a patient clearing the infection (low-range viral load initially, which had cleared by the time the second sample was collected). However, even if the results for all 4 of these indeterminate patients were caused by direct contamination events, the reflex testing rate of contamination from our clinical data would still be low (0.45%; 4 of 894 true negatives). Accounting for indeterminate patients, only 3.04% of all HCV antibody–positive patients (43 of 1,414; Figures 2A and 2B) whose samples were reflex tested were unable to receive a positive or negative test result, indicating that almost 97% of patients in our reflex testing population were able to receive a conclusive HCV test result from submission of a single blood sample.

Figure 2:

Six-month analysis of clinical samples reflex tested for HCV in public health laboratories across Alberta: stratification of (A) HCV antibody specimen outcomes and (B) indeterminate specimens^*

*Duplicate patient specimens: recorded patients with more than one sample tested for HCV in the public health system, corresponding to different specimen collection dates

HCV = Hepatitis C virus; Anti-HCV = HCV antibodies; X̄ = Mean viral load from testing on the cobas^® 6800

Finally, we assessed the efficiency of HCV reflex testing for patients by evaluating mean diagnostic turnaround times over 6-month periods before and after implementing reflex testing in the public health laboratory. Mean turnaround times were only analyzed for first-time HCV antibody–positive patients, which we defined as not previously being positive in the public health system on the basis of a manual look-back of PHN, name, and date of birth. This was because patients with a history of antibody positivity may only receive follow-up testing every 6 months, whereas patients who are first-time positive are immediately prioritized to stage infection and be linked to specialty care (16). In the 6 months before reflex testing implementation (where the two-sample method was used), 983 first-time HCV antibody–positive patients were identified, with an average turnaround time of 39 days (95% CI 34.8 to 42.3 days) between specimen collection for HCV antibody testing and confirmation of HCV RNA results (Table 4). In the 6 months after reflex testing implementation, 1,209 first-time HCV antibody–positive patients were identified, with an average turnaround time of 4 days (95% CI 3.8 to 4.1 days) from time of collection to confirmation of HCV RNA results. Overall, the reflex testing method significantly lowered mean diagnostic turnaround times for patients compared with the standard two-sample method (p < 0.0001) and decreased the minimum number of blood samples required to receive an HCV diagnosis.

Table 4:

Comparison of mean diagnostic turnaround times for first-time HCV positive patients 6 months before and after implementing reflex testing in public health laboratories across Alberta, Canada

Period	HCV antibody–positive patients	Turnaround time, days mean (SE),* 95% CI†, ‡
Before reflex testing implementation (2-sample method)	983	39 (1.911), 34.8 to 42.3
After reflex testing implementation (single-sample method)	1,209	4 (0.067), 3.8 to 4.1

^*Turnaround time represents the period from first specimen collection date to HCV RNA test validation date

† 95% confidence intervals are reported with upper and lower interval values

‡p < 0.0001 (statistical analysis performed using a two-sample t-test with unequal variances; p < 0.05 is significant)

HCV = Hepatitis C virus

Discussion

With the availability of highly effective direct-acting antivirals against HCV (17), the WHO aims to eliminate global viral hepatitis by 2030, with Canadian efforts for elimination driven by the Canadian Network on Hepatitis C (18,19). Improvements in HCV screening and linkage to care are imperative to achieve this goal, particularly because the virus often causes asymptomatic infection. The HCV diagnostic protocol typically requires one blood sample for serology and a second sample for HCV molecular testing. This two-sample method, implemented in acute clinical and public health laboratories around the world (20), is thought to be a major limitation in diagnosing HCV because of the challenge of maintaining patient engagement at the early screening stage (21). Although using a single blood sample for both HCV antibody and molecular testing would be clinically optimal for patient engagement, there have previously been concerns about sample contamination when using one sample across two different testing platforms (12).

Because serology testing platforms using fixed probes may result in upstream contamination, it is possible that instruments with disposable probes may be better equipped for HCV reflex testing strategies. Indeed, a study investigating HCV reflex testing contamination from the cobas 8000 analyzer (which uses single-use disposable tips for serology testing) to the cobas 6800 showed no contamination in 120 samples that were reflex tested (22), suggesting that contamination risk is lower on serology testing instruments without fixed probes. When comparing contamination rates using the same fixed probe serology platform from our study, a Swedish team previously reported a contamination rate of 4% after HCV antibody testing on the Abbott ARCHITECT followed by HCV RNA testing on the cobas AmpliPrep/cobas Taq-Man48 Analyzer (23). Likewise, our HCV reflex testing method (Abbott ARCHITECT to the cobas 6800) showed a similar contamination rate (3.29%), suggesting that contamination rates remain relatively low when using serology testing platforms with fixed probes.

When samples were tested with the cobas 6800, the mean recorded viral loads were overall significantly higher than those tested on the Abbott m2000 (initially used for viral load detection as a dedicated molecular sample), but this was only significant for the samples with a high-range viral load. Regardless, none of our samples had a significant reduction in measured viral load. This is in contrast to the Swedish study, where 32% of HCV RNA–positive samples were shown to have a one-third reduction in viral load after reflex testing compared with original HCV viral load titres (23), suggesting that the cobas 6800 platform may be better adapted for molecular testing than the cobas AmpliPrep/cobas Taq-Man48 Analyzer. In addition, we used samples that represented all available HCV genotypes and ranges of RNA viral loads, and we demonstrated a high sensitivity (99.3%) and specificity (100.0%) of the cobas 6800 assay using reflex testing, with a high level of agreement (99.7%) between the cobas 6800 assay and the standard Abbott m2000 assay. This surpassed the sensitivity and specificity thresholds (≥95%) needed for assays to be implemented in clinical laboratories (14), and as a result, we proceeded to implement this method in public health laboratories across Alberta.

To ensure that reflex testing would be clinically successful, we set the cut-off for an HCV-positive sample at 1,000 IU/mL, 2 standard deviations above our highest-seen contamination value (20.7 IU/mL). Because our indeterminate range was quite broad (1–1,000 IU/mL), we examined the number of patients falling in this range who would subsequently have to submit a second blood sample for follow-up; after implementing reflex testing for 6 months, only 3.04% (43/1,414) of all presumed first-time HCV antibody–positive patients whose samples were reflex tested had indeterminate viral loads. Because our previous data showed that only 70% of patients screening positive for HCV antibodies had documented molecular follow-up testing (11), one concern was that indeterminate patients from our study may likewise not submit a second sample for follow-up testing. However, we saw that the majority of our indeterminate patients (30 of 43; 69.8%) had a record of follow-up testing, and of those, 85% were HCV RNA positive, suggesting that our cut-off was appropriate for our patient population, and follow-up submission rates remained similar with the new process. Although patients with indeterminate results from reflex testing were unable to get a diagnosis with a single blood sample, we were able to show that approximately 97% of patients were able to achieve a conclusive test result from the single-sample reflex testing algorithm.

From a public health perspective, two targets of implementing new HCV testing programs are (1) methods that are efficient on a broad scale and (2) improved patient care and outcomes (24). In our study, we showed that reflex testing significantly improved mean diagnostic turnaround times for first-time HCV antibody–positive patients in Alberta (from 39 d with the two-sample method to 4 d with reflex testing), demonstrating a direct improvement in HCV testing efficiency and further emphasizing the benefits of using the reflex testing algorithm in public health laboratories. A 2016 review from Cadieux et al (25) indicated that the majority of all provinces and territories in Canada did not have any form of reflex testing implemented. A more recent review on HCV elimination from Action Hepatitis Canada (26) showed an increase in the number of provinces and territories with HCV reflex testing, although only half of all provinces and territories currently have an HCV reflex algorithm in place. The public health laboratory at the British Columbia Centre for Disease Control (BCCDC) was among the first to implement reflex testing in its population. It showed that in the 28 years before implementation of reflex testing, 17% of HCV antibody–positive patients in British Columbia never received follow-up molecular testing; however, there are not yet published data on whether reflex testing has improved these numbers (27). In our study, we showed that reflex testing improved molecular testing for HCV antibody–positive patients in Alberta (70% receiving molecular confirmation before reflex testing implementation (11) to almost 97% after implementation). On the basis of these results, it is likely that other provinces would also see an improvement in molecular testing rates after implementation of reflex testing strategies.

Because HCV reflex testing is only used for patients without a history of previous HCV positivity, it is important to consider methods that would be effective at identifying previous positives before undertaking testing. The laboratory in Alberta currently uses a CCL look-back method for previous positivity that relies on in-province PHNs. However, we found an additional 205 previously positive patients over the 6-month period after implementation of reflex testing using a manual look-back method in the LIS compared with the CCL (1,209 patients versus 1,414 patients, respectively; Table 4, Figure 2). This was because our manual look-back method incorporated PHNs, regardless of province origin, and a combination of name and date of birth could be used to trace patients without recorded PHNs. Our data therefore suggest that laboratory resources in Alberta are still being overused for unnecessary antibody testing and that laboratories should consider expanding their criteria for performing look-backs of previous positives before undertaking reflex testing strategies.

Although using a single sample across two testing platforms has previously been a concern for contamination, preventive steps can mitigate contamination risk. Because fixed probes used on the Abbott ARCHITECT instrument can be a source of HCV RNA carry over, laboratory staff should try to minimize splashing when loading the blood sample into the instrument carrier. In addition, laboratory workflow can lead to sample contamination, and it is therefore imperative that molecular-level precautions for serology samples be followed for specimen preparation and aliquoting. Analysis of our laboratory workflow showed a %CV of 4.80% repeatability and 0.57% reproducibility, well below the accepted threshold of variability, suggesting that a standardized, validated workflow can be established to mitigate the risk of contamination and sample variation.

One limitation of our study was the introduction of a single freeze–thaw cycle between using the sample for initial molecular testing and using it for our experiments, which may have caused analyte degradation to occur and could be a source of discordant results. Previous studies have reported upward of 15% HCV viral titre loss after five freeze–thaw cycles (28). Although our reflex viral titres for samples with viral loads of more than 1,000,000 IU/mL were higher than original titres from the Abbott m2000, sample degradation may still be a factor in samples with lower viral load (<1,000,000 IU/mL). However, we saw no significant difference in low-range or mid-range titres from our reflex testing results compared with original titres measured on the Abbott m2000, suggesting that, overall, degradation did not significantly affect the viral titres in our study.

Another limitation of our study was the range of HCV genotypes that were available for use. We aimed to analyze the reflex testing algorithm on all HCV genotypes; however, only specimens with genotypes 1a, 1b, 2, 3, and 4; mixed type; and not typed were available for testing. Genotypes 5 and 6, which are most prevalent in Southeast Asia and South Africa (29) and make up less than 5% of global HCV (30), were not present among specimens collected for the study. Further research encompassing reflex testing on HCV samples with genotypes 5 and 6 should be done before implementation in regions where those genotypes are most abundant. However, 99% of all samples tested at the public health laboratory in 2019 were genotypes 1–4 or mixed type, suggesting that our sample selection was representative of the genotype distribution in Alberta (data not shown). These genotypes are exceedingly present in North America, South America, Australia, and most of Africa and Europe (31); thus, our study supports the use of the cobas 6800 platform for reflex testing where genotypes 1–4 are prevalent. This method could be particularly beneficial in places where HCV prevalence is high, such as Western Africa, Northeast Africa, Russia, and Western Asia (32). Last, although our study showed a significant improvement in turnaround times for diagnostic testing, it is unclear whether this directly improves patient outcomes downstream in the cascade of care, such as being referred to specialists, being prescribed treatment, and achieving a sustained viral response. Our study does, however, support improved patient care at the screening and diagnostic stages, because patients would require fewer health care visits to submit blood samples and would receive test results faster with HCV reflex testing.

Overall, we demonstrated that a single blood draw can be used to diagnose HCV infections accurately and efficiently. Our data support low contamination rates while maintaining high sensitivity and specificity of the molecular assay. Most notably, implementation of HCV reflex testing in the clinical laboratory was shown to improve diagnostic turnaround times for patients and resulted in fewer than 3.05% of patients requiring a second blood sample for confirmation. Broad implementation of single-sample reflex testing algorithms in other public health and clinical laboratories could help improve the screening and diagnosis of HCV within populations. To reach the goal of global hepatitis elimination in the next decade, reflex testing should be considered an important method for identifying HCV infections and improving the efficiency of patient diagnosis.

Funding Statement

Funding for this study was provided in part by the Canadian Institutes of Health Research Frederick Banting and Charles Best Canada Graduate Scholarship (LA Thompson), University of Alberta Doctoral Recruitment Scholarship (LA Thompson), and M.S.I. Foundation (CL Charlton).

Ethics Approval:

This research was approved by the University of Alberta Research Ethics Board (Pro00092635).

Informed Consent:

N/A

Registry and the Registration No. of the Study/Trial:

N/A

Funding:

Funding for this study was provided in part by the Canadian Institutes of Health Research Frederick Banting and Charles Best Canada Graduate Scholarship (LA Thompson), University of Alberta Doctoral Recruitment Scholarship (LA Thompson), and M.S.I. Foundation (CL Charlton).

Disclosures:

The authors have nothing to disclose.

Peer Review:

This manuscript has been peer reviewed.

Animal Studies:

N/A