Analytical aspects of the antinuclear antibody test by HEp-2 indirect immunofluorescence: EFLM report on an international survey

Substrate, conjugate and screening dilution

Substrate heterogeneity can induce variability between results obtained in different laboratories. Comparative studies identified differences between substrates with respect to pattern recognition [2], [3], [4] and sensitivity for particular antibodies (e.g. anti-SSA/Ro60) [5], [6], [7], [8], but results of these studies are ambiguous and probably also biased by differences in conjugates.

In our survey, 69% of the respondents used HEp-2 cells, 16% HEp-2000 cells and 15% HEp-20-10 cells (Table 2). However, geographical differences were observed. For example, 76% (n=155/204) of ICAP respondents used HEp-2 vs. 46% (n=27/59) of Belgian respondents. The use of both the classic HEp-2 substrate and the manipulated cellular substrates (HEp-2000 and HEp-20-10) was supported by the recent EFLM/EASI/ICAP Delphi exercise [92% high scores (≥7 on a scale of 9)] [1].

The isotype specificity of the used conjugate (polyvalent/IgG-specific) may contribute to assay variability (reviewed in [1]). IgG-specific conjugate is considered sufficient to detect most clinically relevant ANA in the context of ANA-associated rheumatic diseases (AARD) [9], [10], [11], [12]. Polyvalent conjugate may produce background fluorescence due to the detection of clinically irrelevant antibodies [9, 10]. Our survey showed that about half of the laboratories (57%) worked with an IgG-specific conjugate, while 37% used a conjugate that detects both IgG heavy and light chains (polyvalent) (Table 2).

Seventy-five percent of the laboratories that participated in the survey utilized combinations of substrate/conjugate as proposed by the manufacturer (commercial kit-combinations). The other laboratories made their own combinations of substrate and conjugate (Table 2).

There is no consensus in literature and amongst experts on the optimal starting dilution to screen (screening dilution) for ANA by HEp-2 IFA [13]. The majority of the respondents of the current international survey (61%) applied a 1:80 screening dilution (Table 2), which is comparable to what was observed in the 2014 EASI survey (60.5%) [14], but lower than the 80% reported by ICAP in 2021 [13]. Dilution 1:160 was used for screening by 15% of the respondents.

Eighty percent of the respondents used the same screening dilution for children and adults. As summarized in the EFLM/EASI/ICAP recommendation paper [1], there is no evidence for screening children for AARD with a lower dilution for ANA by HEp-2 IFA than the dilution used for adults.

Automated microscopy: expert revision, use of FI score for titer estimation and added value in quality assurance

The implementation of CAD represents an important step forward in HEp-2 IFA analysis. The recent EFLM/EASI/ICAP Delphi round documented high support (i) on the added-value of CAD systems for positive/negative interpretation (in combination with expert review) [93% high scores (≥7/9)] and (ii) on the appreciation that expert review remains mandatory for pattern assignment [99% high scores (≥7/9)]. The data of the survey showed that the majority (69%) of laboratories equipped with CAD always reviewed the automatically proposed patterns. In 17% of the laboratories pattern revision was only performed in specific cases. A minority (14%) never reviewed the patterns (Table 3).

Many CADs provide a quantification of the fluorescence intensity (referred to as ‘FI measure’), which significantly correlates with the titer and provides information on the likelihood for AARD [15], [16], [17], [18], [19], [20], [21], [22]. Some systems also provide an estimated endpoint titer based on analysis at screening dilution which is substrate-specific and pattern-dependent [23]. The survey indicated that 49% of respondents using CAD did not use the titer estimation as provided by the system for reporting. About 25% of the laboratories relied on automatic titer estimation for all nuclear patterns and 25.5% for some nuclear patterns (Table 3). Even though >20% of the participants in the EFLM/EASI/ICAP Delphi exercise indicated inexperience on the topic, 87% of the remaining participants supported [high scores (≥7/9)] the idea that the FI measure can be used to estimate the titer and to provide information on the likelihood for AARD [1].

The availability of a reproducible FI measure as provided by the CAD systems brings the HEp-2 IFA to the level of a (semi)quantitative assay and therefore allows for improved evaluation of the quality of the analysis (e.g. quantitative monitoring of within- and between-run reproducibility on internal quality controls (IQC) samples and monitoring of the median FI of a run) (reviewed in [1]) Moreover, the international EFLM/EASI/ICAP recommendation endorsed this approach of quality assessment [92% high scores (≥7/9)] [1]. However, our survey shows that, in daily practice, 53% of laboratories having the FI measure available did not take advantage of these new opportunities (Table 3).

Training, controlling inter-observer variability and external quality control programs

The results of our survey showed that inter-observer variation was controlled in 79% (n=218/277) of labs (Table 4). The approaches applied to control inter-observer variability were double reading (the reported result is a consensus of 2 observers) in 56% of the labs and periodic blinded reading of representative cases more than once a year (formal tests) in 16% of the labs. Some labs combined double reading and periodic blinded reading (10%) or used another approach to control inter-observer variability (18%). The formal tests were mostly organized with a combination of external quality control (EQC) samples and patient samples (68%). The laboratories that did not use both types of samples either used EQC or patient samples (Table 4).

Participating in EQC programs contributes to monitoring the HEp-2 IFA performance (reviewed in [1]). Accreditation programs or national legislation can make participation obligatory. In some countries, as in Belgium, a mandatory national EQC program is organized by national health agencies. Ninety-one percent of the laboratories (n=249/273) participated in EQC schemes, 61% subscribed to minimally 3 EQC distributions a year, and 73% evaluated yearly at least 4 EQC samples (Table 5).

Internal quality control

IQC by monitoring the results obtained with IQC samples in each run is one of the most important cornerstones in the quality control (QC) program of clinical laboratories. Our survey shows that IQC for HEp-2 IFA was performed by 96% of the laboratories (Table 6).

Many guidelines support the need for inclusion of at least 2 IQC samples (one negative and one low positive) in each run and advice semi-quantitatively judgement of the results (either by end-point titration or automated intensity scoring) [1, 9, 24, 25]. The results of our survey indicate that 78% of laboratories indeed evaluated at least 1 positive and 1 negative control sample in every run. About 23% of the laboratories only analyzed at least 1 positive control but no negative control. Less than a fifth of the laboratories (18%) used more than 1 positive sample as run IQC. About half of the laboratories positioned all their IQC samples together at the start or end of the run; the others spread the IQC samples with some at the start and some at the end (Table 6).

The EFLM/EASI/ICAP recommendation states that the positive control should be a low positive sample matching an LR of 2–5 for ANA-associated rheumatic diseases [1]. The survey revealed that positive samples had mostly high (52%) or medium (19%) titers. Cut-off controls were only used in 13% of the laboratories (Table 6).

Different approaches can be used for choosing the pattern(s) of the positive IQC sample(s). The 2009 German EASI recommendation suggests testing IQC samples with different patterns, in alternation over the runs [26]. Another approach focuses on the sensitivity for fragile antigens, such as SSA/Ro-60, and proposes to include antibodies to such antigens as IQC [6]. Still another approach is to select the patterns based on high inter-assay reproducibility of the FI measure on CAD systems, allowing optimized IQC approaches ([27], [28], [29], [30], [31], reviewed in [1]). This approach focussing on selecting a pattern with high reproducibility for IQC purposes obtained high international support during the EFLM/EASI/ICAP Delphi exercise [98% high scores ≥7/9 with median score 8] [1].

The survey revealed that of the labs that perform IQC, most (62%) used always the same pattern over the different runs. In those labs that alternate the patterns, the frequency of alternating the patterns differed between labs (Table 6). When labs were running more than 1 positive IQC per run, 83% (n=39/47) tested different patterns. Fifty-two percent of these labs tested at least 3 different antigen specificities/patterns per run.

Controls included in commercial kits are often pre-diluted and mostly have high titres. This precludes detection of pipetting errors. Therefore, combining IQC samples from patient origin with commercial controls is advised ([24, 25, 27, 28], reviewed in [1]). In our survey, 70% of laboratories used undiluted IQC samples or IQC samples that were differently diluted than the patient samples. Many labs (58%, n=172/296) exclusively used commercial IQC material. Thirteen percent of the respondents relied only on control material from patient origin, and almost one-third (29%) of the labs combined both. When using patient samples as IQC material, most laboratories (64%) used single patient samples rather than pools, while 24% used both single patient samples and pools. Laboratories using exclusively pools were a minority (11%). Seventy-five percent (n=191/256) of laboratories using commercial QC material adopted the company target, and 19% determined a laboratory target (see Table 6).

CLSI suggests long-term storage of samples at −70 °C in combination with a preservative [9]. It is however common practice to freeze samples at −20 °C, as reported by 71% (n=90/127) of the labs participating in the survey. Ten percent of the laboratories preserved patient QC material at −80 °C (Table 6). Twenty-nine percent of the laboratories used the fridge for longer than 1-week storage, contrary to the CLSI guidelines that advises limiting preservation in the fridge to 3 days [9, 32]. Only 7% safeguarded the samples with preservatives.

Quality control approaches based on patient results

Monitoring the percentage of (low, medium, high) positive patient results in Levey-Jennings plots allows us to evaluate assay stability over time [9, 28, 29]. Such an approach is considered to be of added-value, as reflected by the high score (≥7/9) given by 87% of the experts participating in the EFLM/EASI/ICAP Delphi exercise [1]. In the survey, however, only 17% (n=51/292) of the respondents monitored the percentage of positive patients.

Reagent lot acceptance and monitoring of lot-to-lot variability

Variations in HEp-2 IFA results have been shown between different slide brands [2], [3], [4] and suggested to be present also between reagents lots [2, 29, 31, 33]. According to international guidelines and EN/ISO 15189:2012 accreditation requirements, lot-to-lot variability of different batches (of one brand/kit) should be evaluated before implementing a new lot [24,34]. According to the EFLM/EASI/ICAP recommendation this can be done by analysis of patient-derived IQC samples supplemented with samples selected for this purpose ideally covering different cell compartments (nucleus and cytoplasm) and different titer levels [1]. Our survey revealed that lot-to-lot evaluation of substrate was performed by 68% (n=173/254) of the laboratories. This was mainly performed with commercial samples, either exclusively (42%) (n=72/173) or combined with patient material (40%) (n=69/173) (Table 7). Eighteen percent (n=31/173) of laboratories used solely patient samples. Approximately one third (37%, n=33/89) of the laboratories evaluating patient samples tested only 1 pattern. Two or 3 patterns were tested by 24% of the laboratories and more than 3 patterns by 30% of the laboratories. The EFLM/EASI/ICAP recommendation suggests testing with samples reacting with different cell compartments (nucleus and cytoplasm) of titer levels including negatives [1]. To limit the variation linked to lot variability, 47% of the laboratories ordered reagents of 1 lot for a longer period (Table 7).