Reporting of Antinuclear Antibody HEp-2 Indirect Immunouorescence Assay: A Performance Survey

Background: To evaluate the interpretation and reporting of antinuclear antibodies (ANA) by indirect immunouorescence assay (IFA) using HEp-2 substrate based on common practice and guidance by the International Consensus on ANA patterns (ICAP). Method: Participants included two groups [16 clinical laboratories (CL) and 8 in vitro diagnostic manufacturers (IVD)] recruited via an email sent to the Association of Medical Laboratory Immunologists (AMLI) membership. Twelve (n=12) pre-qualied specimens were distributed to participants for testing, interpretation and reporting HEp-2 IFA. Results obtained were analyzed for accuracy with the intended and consensus response for three main categorical patterns (nuclear, cytoplasmic and mitotic), common patterns and ICAP report nomenclatures. The distributions of antibody titers of specimens were also compared. Results: Laboratories differed in the categorical patterns reported; 8 reporting all patterns, 3 reporting only nuclear patterns and 5 reporting nuclear patterns with various combinations of other patterns. For all participants, accuracy with the intended response for the categorical nuclear pattern was excellent at 99% [95% condence interval (CI): 97-100%] compared to 78% [95% CI: 67-88%] for the cytoplasmic, and 93% [95% CI: 86%-100%] for mitotic patterns. The accuracy was 13% greater for the common nomenclature [87%, 95% CI: 82-90%] compared to the ICAP nomenclature [74%, 95% CI: 68-79%] for all participants. Participants reporting all three main categories better performances compared to those reporting 2 less The average accuracies varied between participant groups, however, with the lowest and most variable performances cytoplasmic pattern specimens. The reported titers for all specimens varied, with the least variability nuclear patterns and most titer variability associated with cytoplasmic patterns. Conclusions: Our study


Background
The presence of antinuclear antibodies (ANA) is a hallmark and classi cation criterion for a number of systemic autoimmune rheumatic diseases (SARD). ANA testing is usually performed as part of the initial diagnostic workup when suspicion of an underlying autoimmune disorder is high. The indirect immuno uorescence antibody (IFA) technique on HEp-2 substrate has been considered the traditional and preferred method for detecting ANA by some [1]. It allows detection of antibody binding to speci c intracellular targets, resulting in diverse staining patterns that are usually categorized based on the cellular components recognized and the degree of binding, as re ected by the uorescence intensity or titer [2,3]. As a screening tool, the recognition of a well-de ned HEp-2 IFA staining pattern may be helpful in determining the most likely speci c autoantibodies present, as well as suggesting possible clinical associations for known speci cities [3,4]. In this regard, a positive HEp-2 IFA screening pattern can guide con rmatory testing and may also be useful for elucidating a speci c clinical diagnosis or prognosis. Thus, the provision of the HEp-2 IFA pattern and titer is considered to be clinically valuable with favorable utility in comparison with other methods for ANA detection [1][2][3][4][5][6][7][8][9][10][11].
The nuclear IFA staining patterns most commonly recognized and reported by clinical laboratories include homogeneous, speckled, centromere, and nucleolar [1][2][3][4][12][13][14]. Use of HEp-2 cell substrates, permits detection of additional nuclear staining patterns, as well as reactivity with cell constituents in compartments outside the nucleus (cytoplasmic) and cell components associated with mitosis (mitotic) [2,4,[12][13][14]. However, the reactivity and type of autoantigens associated with these patterns may vary among HEp-2 substrates from different manufacturers [15]. Furthermore, the expertise required to identify the different patterns and sub-classify their variants may not be universally available in clinical laboratories. Traditional legacy recommendations for reporting ANA patterns on HEp-2 cells continue to signi cantly in uence clinical laboratory reporting [16,17].
The rst International Consensus on Antinuclear Antibody (ANA) Patterns (ICAP) was published in 2015 to systematize and update reporting of autoantibody patterns detected by IFA using HEp-2 cell substrates [12]. The goal of this initial and subsequent publications was to optimize usage of HEp-2 IFA patterns in patient care, by promoting standardization, harmonization and understanding of autoantibody test nomenclature and providing guidelines for test interpretation and reporting [4,[12][13][14][15]18]. To-date, 30 HEp-2 IFA nuclear, cytoplasmic and mitotic patterns have been elucidated by ICAP and presented in a classi cation tree [19]. The ICAP guidelines indicate that 'expert-level' laboratories would report all the HEp-2 IFA patterns, whereas those designated as 'competent-level' laboratories would report 6 nuclear and 5 cytoplasmic HEp-2 IFA patterns [12].
In a previously reported survey administered in cooperation with the Association of Medical Laboratory Immunologists (AMLI), a signi cant number of respondents were unaware of the ICAP initiative, although a majority agreed on the need to standardize the nomenclature and reporting of HEp-2 IFA results [20]. Based on the responses from this survey, a consensus to improve ICAP awareness and further enhance HEp-2 IFA assessment through increased collaboration between ICAP and the clinical laboratory community was suggested with emphasis on education and availability of reference materials. As others have also reported [21][22], many laboratories around the world are inclined to adopt the ICAP nomenclature and embrace the recommendations provided in these consensus guidelines.
The objective of this study was to evaluate the performance of HEp-2 IFA interpretation based on nuclear, cytoplasmic and mitotic staining in an endeavor to characterize competency as outlined in the ICAP classi cation.

Participants and recruitment
Participants for the survey were recruited via an email sent to the Association of Medical Laboratory Immunologists (AMLI) membership (a professional organization focused on immunological laboratory testing with 105 active members) and 5 in vitro diagnostics manufacturers (IVD). The requirements to participate included testing pre-de ned specimens by IFA using HEp-2 substrate and reporting the patterns observed based on speci c "traditional" (as de ned below) as well as ICAP [19] nomenclature for reading and reporting ANA patterns. Sixteen (n = 16) clinical laboratories (CL) and all 5 in vitro diagnostics manufacturers (IVD) agreed to participate in the study. Three additional IVD (2 in the US and 1 in Europe), contacted the organizers to participate in the survey. Overall, 16 CL and 8 IVD participated in this performance survey (Additional File 1).

Study specimens and survey
Twelve (n = 12) specimens were used in the survey. The specimens were chosen based on an assessment of need (described below) by some of the authors (AET, LKP, EKLC, MJF, MHW). The following attributes were taken into consideration: 1) the three main categorical group of HEp-2 IFA patterns, 2) the ICAP guidance for both competent and expert levels, 3) the clinical signi cance of patterns and 4) whether or not pro ciency testing was available for speci c patterns. The authors also wanted to evaluate how participants would interpret nuclear staining associated with anti-topoisomerase I antibodies given recent ICAP guidance for evaluating this complex pattern [23]. The twelve pre-speci ed HEp-2 IFA-positive specimens labeled ANA-001 through ANA-012 included those positive for nuclear [ANA-002, ANA-003, ANA-005, ANA-006, ANA-007, ANA-010, ANA-011], cytoplasmic [ANA-004, ANA-008, ANA-009] and mitotic [ANA-001, ANA-012] categorical groups of IFA patterns (Table 1). All specimens for the survey and their intended responses were obtained from Plasma Services Group Inc. (PSG: Huntington Valley, PA, USA). Specimens were quali ed at PSG using routinely available methods and also veri ed in the laboratories of one or more expert members of ICAP (PSG, personal communication). The complex pattern involves speckled nuclear staining, and also includes staining of the condensed chromatin, cytoplasmic staining, staining of the nucleolar organizing region in mitotic cells, and variable nucleolar staining of interphase cells. **For this specimen, AC-25 was also considered acceptable.
Survey specimens were shipped to all participants in January 2020 with detailed instructions for testing as well as a report form to record and return results to one of the organizers (AET). Parameters to be recorded by checking the survey form included the three categorical groups of HEp-2 IFA patterns reported (nuclear, cytoplasmic or mitotic); commonly used nomenclature (also referred to as traditional in this investigation) for 5 HEp-2 IFA nuclear patterns (homogeneous, speckled, centromere, nucleolar, discrete nuclear dots), mitotic, cytoplasmic or 'other' in accord with legacy classi cation approaches [16,17]; and a result based on the ICAP classi cation tree [19], which includes more detailed sub-pattern classi cation than commonly reported. In addition, the participants were requested to provide information about how the images were read and interpreted (manual and/or automation-assisted reading); the years of experience of the reading technologist(s); the manufacturer of the HEp-2 substrate; the laboratory's typical practice about reporting only nuclear patterns vs also reporting cytoplasmic and/or mitotic patterns when the ANA test is requested; the screening dilution(s) of serum used for detection of ANA in performing the HEp-2 IFA; and the titer of the ANA, based on serial dilution of the tested specimen. There were two types of participants: clinical laboratories (CL) and in vitro diagnostic manufacturers (IVD). After the results were tabulated, participants were not afforded the opportunity to adjust or revise responses based on the responses of other respondents. Some of the participating CL included those directed by the authors, but the authors did not participate in assigning the patterns reported from their laboratories.

Data analyses
We compared participants' HEp-2 IFA pattern classi cation of specimens against a consensus classi cation. The primary outcome was the percent accuracy between the participant and consensus classi cation. We studied the impact of three factors on accuracy: classi cation nomenclature (N), organization type (T), and experience (E). We examined three nomenclatures: 1) group category (nuclear, cytoplasmic, mitotic); 2) speci c traditional pattern descriptions (e.g. speckled, nucleolar, etc); and 3) subpattern classi cation using the ICAP nomenclature. We refer to these as the group, traditional and ICAP classi cation methods. Each participant was classi ed according to the organizational type and the reporting experience at their institution. There were two types of organizations: 1) clinical laboratories (CL) and 2) in-vitro diagnostic manufacturers (IVD). Organizations were classi ed as experienced if they routinely report all group categories (E = yes) and inexperienced (E = no) if they do not routinely report the three group categories. Overall, there were three components, N, T and E: N had three levels (group, traditional, ICAP), T had two levels (CL and IVD) and E had two levels (yes, no). We sought to answer the following questions: 1. Was accuracy associated with the classi cation method?
2. Was accuracy associated with experience?
3. Was accuracy associated with organization type among experienced participants?
4. Was accuracy associated with categories within nomenclature methods?
We used logistic regression to determine the association between accuracy and the three factors: N, T and E. Outcomes were reported as odds ratios (OR). P-values were adjusted for multiple comparisons using the method of Holm. Statistical analyses were performed using Stata 16.2 (Stata Corp LLP).

Characteristics of survey participants
There were 24 participants: 16 were CL (13 in the United States and 3 in Canada) and 8 IVD. Most of the CL used kits from three main IVD that also participated in the survey ( Table 2). The majority of the CL read, interpreted and determined HEp-2 IFA patterns and titers manually using 1:40 as cut-off for HEp-2 IFA determinations. The CL using the Inova automated image reader used the 1:80 cut-off (as recommended by the manufacturer). The median years of experience for technologists who participated in the survey was 10 years for CL compared to 20 years for the IVD participants. manufacturers, n = 9) was involved in the reading and interpretation of the results. ***One laboratory in each group reports cytoplasmic pattern only as a comment.
The number of categorical groups of patterns typically reported by the CL was variable. Among the 16 CL, 3 indicated they reported only nuclear patterns, 3 indicated they reported nuclear and mitotic patterns, 2 reported nuclear and cytoplasmic patterns (with one of the 2 reporting the cytoplasmic pattern as a comment, not as a 'positive ANA') and 8 indicated the laboratory reported nuclear, cytoplasmic and mitotic patterns (with one of the 8 reporting the cytoplasmic patterns as a comment, not as a 'positive ANA'). Among the 8 IVD, 6 indicated they report patterns in all 3 categorical groups, however, all provided responses to all categories in the survey.

Performance of participants based on "traditional" and ICAP nomenclatures
Participants were asked to report results based on survey suggested classi cation (referred here as "traditional") as well as the ICAP nomenclature. The overall accuracy for the traditional nomenclature system was 87% (95% CI: 82-90%), Table 4. Only the specimen (ANA-011) with antibodies to DNA topoisomerase I was reported with accuracy less than 80% and that complex mixed/compound pattern had not been included in many ANA pattern classi cation teaching schemes prior to its recent inclusion as a distinct ICAP pattern [20]. Among the traditional pattern reports, two specimens (ANA-003, centromere and ANA-006, nucleolar) were reported with accuracy of 100%. Legend: ICAP = International Consensus on Antinuclear Antibody Patterns). ID: identi cation number; AC: anti-cell; DND: discrete nuclear dots; MND: multiple nuclear dots; CS: coarse speckled; DFS: dense ne speckled; FS: ne speckled; AMA: anti-mitochondrial antibodies; homo: homogeneous; anti-topo I: anti-DNA topoisomerase I; CI: con dence interval; NuMA: nuclear mitotic apparatus. *Numbers in table re ect the fact that not all laboratories reported cytoplasmic or mitotic categorical group patterns, and **nomenclatures for some specimen.
Several specimens yielded unexpected results. For example, ANA-010 was intended to represent a nuclear ne speckled pattern (AC-4), but that specimen was reported as ICAP pattern AC-4 by only a minority of CL and IVD. The survey showed that all participants reported this specimen as having a speckled nuclear pattern using traditional descriptions, but a majority (75% of IVDs and 55% of CL) reported it as having coarse speckled nuclear staining (AC-5) rather than the expected AC-4. Review of images from several participating laboratories revealed that the specimen produced patterns ranging from typical ne speckled to coarse speckled nuclear staining using different HEp-2 cell sources. Similarly, the specimen (ANA-009) intended to represent a cytoplasmic ne speckled pattern (AC-20) was reported with other patterns by a majority of participants and review of images from different laboratories showed a variable pattern of staining depending on the source of the HEp-2 substrate (data not shown).
The specimen with antibodies to topoisomerase I (topo-1, AC-29 pattern) was reported as homogeneous by most (56%) CL participants and as a mixed (homogeneous and nucleolar) pattern by an additional 19% of CL using the common pattern descriptions. The IVD participants reported it as having a variety of mixed nucleolar patterns. Using ICAP nomenclature, 88% of IVD participants and 54% of responding CL participants assigned the specimen as having the AC-29 (anti-topoisomerase I) pattern.
Overall, for all participants, the accuracy was 13% greater for the traditional nomenclature (87%, 95% CI: 82-90%) compared to the ICAP nomenclature (74%, 95% CI: 68-79%, Additional File 2B and Tables 3). However, the accuracy for reporting the ICAP nomenclatures were lower for CL than IVD with an overall accuracy of 81% (95% CI: 77-84%, Supplementary Table 2). The accuracy of classi cation was associated with participant type ( p < 0.0005) and nomenclature system (p < 0.0005). The accuracy of the CL group was 15% less than the IVD group.
To assess the performance of each organization type based on the accuracies for the main categorical, traditional and ICAP nomenclature determinations, the data was strati ed, and frequencies of the correct intended responses estimated (data not shown). Both groups were effective in determining the intended nuclear staining, however, the CL group demonstrated lower frequencies of the expected responses for the different nomenclatures. This was most pronounced for the ICAP nomenclature.

Impact of nomenclature and participant experience on accuracy
Participants had the highest accuracy using the group category nomenclature ( Table 3). The average accuracy associated with the group category nomenclature was 93% (95% CI: 90-96%). For the traditional χ 2 1 2 nomenclature, the average was 87% (95% CI: 83-91%) which was signi cantly less accurate than the group category nomenclature (OR = 0.48, p = 0.014). The average accuracy of the ICAP nomenclature was 74% (95% CI: 68-79%) which was signi cantly less the group category nomenclature (OR = 0.20, p = 0.002) and the traditional nomenclature (OR = 0.42, p = 0.002), Table 5. Experienced participants had higher accuracy than nonexperienced participants (OR = 2.2, p < 0.0005). The accuracy of experienced participants was greater than the accuracy of nonexperienced participants for all nomenclatures. The difference was 6% for the group method, 10% for the traditional nomenclature and 13% for the ICAP nomenclature.

Frequency distribution of end-point titers
For each of the 12 specimens analyzed, the distribution of the reported antibody titer was recorded. The screening titer of determinations ranged from 1:10 to 1:80. The titers were generally quite variable (Fig. 1, shown for CL). The specimens with cytoplasmic patterns were often not titered, particularly by laboratories that did not routinely report cytoplasmic patterns. Of these specimens, the titer variability was most pronounced for ANA-004 (AMA, AC-21) with positive results demonstrating a bimodal response which spanned 8 2-fold titers ranging from 1:80 to 1:10:240 for CL reporting this pattern. Nuclear pattern staining titers also varied substantially, spanning from 4 to 7 2-fold titers in different specimens.

Discussion
Using pre-tested and selected patient serum specimens, we report here the performance of 24 CL and IVD participants recruited from a professional organization focused on immunological laboratory testing and experienced in the interpretation and reporting of HEp-2 IFA patterns. The specimens included examples from all 3 main categorical patterns (nuclear, cytoplasmic, or mitotic), using 'traditional' and ICAP nomenclatures and included patterns designated by ICAP as associated with both 'competent' and 'expert' laboratories. Our data demonstrates competence for participants in identifying and reporting common nuclear ANA patterns, but inconsistency in the decision to report and pattern reporting of cytoplasmic and mitotic patterns.
In recent years, efforts to standardize interpretation and reporting of HEp-2 patterns have led to a consensus nomenclature presented by ICAP, a group of experts [4,[12][13][14] with the purpose of systematic reporting and optimizing the usage of HEp-2 IFA patterns in patient care [4]. In a previous study, we identi ed increasing awareness of this guidance; availability of reference materials for training and collaboration between professional organizations, IVD and CL amongst others as key elements necessary for improved harmonization of the HEp-2 IFA reporting [20].
The term "antinuclear antibody'" is commonly used to refer to the HEp-2 IFA test for detection of antibodies to nuclear antigens. However, ICAP intentionally included standardized reporting of autoantibodies to HEp-2 cell compartments other than the nucleus, since some of the cytoplasmic and mitotic patterns are associated with systemic and organ-speci c autoimmune diseases [2,4,12]. Thus, there has been a consensus suggestion to refer to ANA as anti-cellular antibodies [2] and, indeed, ICAP utilizes the anticellular (AC) nomenclature in their pattern classi cations [4]. Specimens with nuclear patterns made up the majority (58%) of the survey specimens, followed by the less commonly reported cytoplasmic (25%) and mitotic (17%) patterns. Of interest, relatively more CL in this survey indicated they reported mitotic than the cytoplasmic patterns, perhaps because some mitotic patterns such as the distinctive spindle pattern (AC-25, and AC-26) have been traditionally taught in conjunction with nuclear staining patterns. Amongst the responders, mitotic patterns were more accurately reported than the cytoplasmic patterns. While this survey may be limited by the number of participants, this observation is in contrast to the current expert-level classi cation by ICAP of all mitotic patterns, with implications for validation in a larger group of participants. On the other hand, except for the AMA (AC-21) pattern, the two other specimens with cytoplasmic patterns, which are intended to be of expert-level, did not achieve consensus. Cytoplasmic patterns are associated with diverse subtypes and different antigenic targets, many with de ned clinical implications. For example, the AMA pattern is typically associated with primary biliary cholangitis (PBC), the dense ne speckled cytoplasmic pattern (AC-19) with anti-ribosomal P in SLE patients and the ne speckled pattern (AC-20) with some anti-tRNA synthetase antibodies (Jo-1) in patients with myositis [4,[24][25][26][27]. Other cytoplasmic sub-patterns include rods and rings associated with Hepatitis C infection treated with ribavarin and the cytoplasmic discrete dots pattern (AC-18) associated with GW bodies described in patients Sjögren's syndrome,neurological disease and other conditions [4,28,29].
In addition to accurately reporting binding of autoantibodies to de ned cellular components, the survey also evaluated responses based on "traditional" categorization for nuclear patterns as well as the emerging ICAP nomenclature. As expected, all participants performed better with the more widely used or common HEp-2 IFA nomenclature, which has more emphasis on limited nuclear staining features than required to correctly assign ICAP patterns. While the reason for this could be due to limited familiarity with ICAP, based on the data, other reasons for this can be inferred. First, the "traditional" categorization which can also be referred to as the ICAP "competent-level" is broad and minimizes the use of ne details and/or integrated pattern recognition in its interpretation. For example, most responders were capable of identifying ANA-003 and ANA-007 as speckled but failed to accurately demonstrate the intended ICAP nomenclatures, AC-5, and AC-2, respectively. In fact, a number of respondents classi ed the AC-2 DFS specimen as a mixed nuclear homogeneous and nuclear speckled pattern, as might be expected for traditional classi cation based on speckled staining of the nucleoplasm and intense chromatin staining. The combination requires integration to assign the nuclear DFS AC-2 ICAP pattern, rather than describing mixed homogeneous and speckled staining pattern with which it might be confused.
The specimen with antibodies to DNA topoisomerase I and the AC-29 staining pattern also demonstrated remarkable challenges of consistent ANA pattern reporting. Under ICAP, AC-29 is considered a sub-pattern of nuclear speckled staining [23], but only a minority of participants reported it as a nuclear speckled pattern using traditional nomenclature. Using "traditional" classi cations, the pattern is a compound, mixed staining pattern in which the speckled component may not be perceived as dominant, even though it is consistently observed. In addition to the speckled nuclear staining, there is also staining of condensed chromatin in the mitotic cells, making it di cult to distinguish from homogeneous nuclear staining, as reported by majority of the CL, and nucleolar staining also is often present. Dellavance and colleagues [30] rst reported on a composite of ve unique attributes associated with positivity for anti-topoisomerase I and HEp-2 IFA pattern which may not be consistently observed in all HEp-2 substrates and/or serum dilutions [23]. Among the value of the ICAP classi cation scheme is that interpretation of complex mixed staining may be better reported as a single unifying pattern. In support of this, laboratories accustomed to ICAP classi cation correctly reported the ICAP AC-29 topoisomerase pattern when asked to use the ICAP nomenclature, although they may not have reported it as a speckled ANA using traditional descriptions.
A recent multicenter analysis to evaluate the interpretation of HEp-2 IFA reported signi cant differences among laboratories in terms of qualitative results, patterns, and titers, particularly at low levels and in those with speckled patterns [31]. HEp-2 IFA titer determinations have been reported to have clinical signi cance in predicting risk for disease (healthy vs. disease) as well as association with speci c autoantibodies [32][33][34][35][36]. Our data con rm previous reports that ANA titers vary considerably, and point out another opportunity for harmonization of ANA reporting. With respect to the ICAP nomenclature, our data demonstrated clusters of participants based on the HEp-2 patterns reported by the participants. First, the majority of participants in this survey reliably read, and interpreted the centromere, multiple nuclear dots, nuclear dense ne speckled, AMA and NuMA-like sub-patterns. The NuMA-like pattern is considered uncommon, and expected to be recognized by "Expert" level laboratories, but it has a characteristic appearance, and has clinically signi cant associations with a number of SARD, as reported in a large cohort of Columbian patients [37].
Second, a signi cant group of participants could identify challenging ICAP-designated sub-patterns. These include the homogeneous nucleolar, cytoplasmic dense ne speckled, spindle ber, nuclear coarse speckled, and anti-topoisomerase I patterns. Except for the AMA pattern, the overall performance of the CL participants for specimens with the cytoplasmic patterns was relatively more variable, and lower than IVD group. These observations have implications for de ning competency for CL for cytoplasmic and mitotic patterns.
A minority of participants interpreted the nuclear ne speckled, and cytoplasmic ne speckled sub-pattern specimens as intended. The data suggested that the HEp-2 patterns generated by those specimens had a su ciently variable appearance, based on the kit manufacturer, and/or kit lot, to lead the specimens to appear as different ICAP categories in the hands of different participants. That hypothesis was con rmed by our direct review of the appearance from different laboratories (data not shown). The observations reinforce the need for harmonization of reagents, as well an enhanced training in pattern interpretation, in order to generate consistent results.
Analyses of the performance of the participants showed that the average accuracy with the expected patterns varied based on the hierarchical categories and rater groups (CL vs IVD). Combined, both group of participants exceeded 80% average accuracy for two (nuclear and mitotic) of three group categorical patterns. The performance for both groups was more variable based on the two different nomenclatures, traditional and ICAP. However, the CL group had more varied average accuracy for both the traditional and ICAP nomenclatures with the ICAP nomenclature demonstrating signi cantly lower performnace. This may re ect how HEp-2 IFA patterns are reported in the CL and/or the experience of these participants. Notably, the participants in the IVD group had more years of experience than those on the CL group. Furthermore, only half of the CL participants reported results for all three group categories. When the results were strati ed based on whether or not all three group categories are reported by the CL group, the performance for this subset was comparable to the IVD group. Based on this observation, it is likely that a signi cant majority of participants that report all three group categories may have developed competencies for the more challenging (expert-level) patterns. It is also likely that the IVD group may not be subjected to regulatory constraints like the CL participants and had increased exibilities in reporting for this survey.
The ICAP classi cation tree indicates patterns that should be readily recognized (competent-level) versus patterns that would be more challenging and/or infrequent and distinguishable only when observers or technologists have attained the expert-level [12]. The specimens chosen for this survey were chosen to cover a range of levels of expertise.
The ICAP guidance is recognized as a potential roadmap towards the harmonization and standardization of HEp-2 IFA nomenclature [38,39]. It is understood by its members and opinion leaders that this guidance will evolve, taking into consideration practical aspects for its adoption in clinical laboratories; diverse experience, ageing workforce, variability in reagents, microscopy and recent introduction of digital image readers [14,20,38]. Along these lines, this investigation is not without limitations. First, the intended responses (traditional nomenclature) for specimens with the cytoplasmic and mitotic patterns were not de ned for speci c sub-patterns (for example, cytoplasmic speckled or NuMA). This was intentional as it was largely unknown how CL report both patterns. The results obtained from this survey validates the approach, as the minority of laboratories reporting less than 3 main categorical patterns do report mitotic pattern considered expert-level on the ICAP classi cation tree [www.anapatterns.org , 12]. Second, the intended responses were monospeci c and did not take mixed patterns into consideration. A number of participants reported mixed patterns for some of the specimens (data not shown), often with the intended dominant pattern reported together with minor additional pattern variants. Such reports were considered appropriate and in accordance for reporting patient results with more than one pattern [2]. Third, the survey included a limited number of participating CL including those with a signi cant interest and experience in ANA testing, which may not re ect the experience of a wider spectrum of international CL. Finally, some of the participants, particularly those in CL group, may have limited familiarity with the ICAP nomenclature, despite being associated with experienced laboratories.
The data presented con rm that standardization of reporting has not been achieved in performance of non-traditional HEp-2 patterns even by experienced and interested laboratories. This suggests the need and opportunities for further training and consensus-building. Using the ICAP nomenclature may have bene ts for some sub-patterns and assigning competencies, notably for the mitotic and cytoplasmic main categorical groups and our data clearly demonstrate that recognition of the pattern associated with antibodies to topoisomerase is linked to familiarity with ICAP patterns. Furthermore, our data con rm previous observations that differences in the HEp-2 cell substrate can contribute to inconsistency in ANA sub-patterns interpretation and reporting [23]. Clearly, consistent ICAP sub-pattern reporting by laboratories is most meaningful if patterns are commutable using different sources of HEp-2 reagents. The relatively higher competencies of the IVD participants relative to the CL participants is of interest as some laboratories depend on IVD for training as gleaned from AMLI practice survey [20].

Conclusion
This study highlights signi cant competency for all participants in identifying the nuclear main categorical HEp-2 IFA patterns. This observation validates the ICAP competent-level classi cation for this group except for the anti-topoisomerase I antibody pattern, with implications for downstream con rmatory testing. Our data also demonstrate opportunities for de ning competencies and training for CL personnel in recognition of cytoplasmic and mitotic patterns.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.