Evaluation of a worldwide EQA scheme for complex clonality analysis of clinical lymphoproliferative cases demonstrates a learning effect

Clonality testing is widely accepted as a valuable tool in routine diagnosis of lymphoid malignancies [1, 2]. The vast majority of lymphoid malignancies arise from the unconstrained expansion of a single transformed B- or T-cell, accompanied by the presence of clonal rearrangements of immunoglobulin (IG) or T-cell receptor (TR) genes, rendering them the most widely applied gene targets for clonality testing.

In 2003, the European BIOMED-2 consortium, continued under the name “EuroClonality consortium,” designed a standardized multiplex PCR assay for nearly all IG/TR targets [3], which showed a high rate of clonality detection in common B- and T-cell malignancies [4‐12]. To date, several commercial kits are available to run the multiplex PCR assays.

Due to the technical standardization and commercialization (PCRs, protocols, and readouts), clonality assays can be performed in routine diagnostics [13]. However, reporting of clonality assays is still considered a complex task, because molecular clonality testing reflects immunobiology and comprises the integrated interpretation of multiple multiplex PCR results. These multiplex PCRs use primers of potentially different efficiencies, annealing to highly homologous genes. Although there are basic rules for interpretation of the molecular patterns [13], extensive knowledge of IG and TR gene rearrangement patterns and the PCR design is needed. Also, interpretation should consider the pathology and the clinical question as the presence of a small clone in a reactive lesion has another implication than its presence in a full-blown malignancy.

Laboratories performing molecular pathology tests are advised to participate in external quality assessment (EQA) schemes [14], preferably an accredited scheme providing samples mimicking routine cases as closely as possible [15]. It is essential that the EQA participants read the final reports with feedback on errors made by all participants, act on recommendations made, and ensure that their own errors are corrected rapidly [14].

The EuroClonality consortium organized five EQA rounds between 2008 and 2011 [16], using capillary electrophoresis (GeneScan, GS) or polyacrylamide heteroduplex (HD) gel analysis [3]. The schemes aimed to (i) assess the laboratory performance and (ii) develop a uniform scoring system for interpretation of IG/TR clonality testing. To render interpretation less subjective, algorithms have been introduced, especially in the USA [17‐19]. However, this may potentially lead to false negative or false positive interpretation [13], and the need for guidelines on interpretation and reporting of clonality data is apparent for IG/TR routine diagnostics and EQA schemes. This prompted the development of the EuroClonality (BIOMED-2) guidelines. The guidelines describe the technical scoring of the individual IG and TR PCR target results, and scoring of the final molecular conclusion, based on the integration of the different PCR results. During validation of the EuroClonality uniform description and reporting system, the majority of the cases were scored appropriately, with only 3.1% of 1150 cases being identified as difficult to score, i.e., the final scoring of either a minor clone with polyclonal background or polyclonal with a minor background, actually describe the same phenomenon but the scoring may reflect the personal favor of the clinical scientist [13].

Several other providers offer IG/TR EQA schemes [20‐24]. Both small sample sets with frequent distribution and larger sample sets are currently used in different EQA programs [25, 26] (supplemental Table 1). Previously, performance improvement upon EQA participation or other quality improvement projects has been reported in schemes for testing of oncological biomarkers [27, 28], but not yet for clonality testing.

The aim of this paper was to investigate the effect of repeated EQA participation on the laboratories’ performance for complex clonality analysis. Important parameters such as the participant group, the different final molecular interpretations (clonal, polyclonal, oligoclonal; without evaluation of the more detailed molecular interpretation), sample information, and the analysis method were integrated in these analyses. The data are based on the results of five EQA rounds for IG and TR rearrangement analysis in suspected lymphoproliferations between 2014 and 2018.

The BIOMED-2/EuroClonality assays are widely used for clonality testing of suspected lymphoproliferations. Clonality testing is not a stand-alone test but is an important integral part in the diagnosis of lymphoid malignancies. Correct analysis, evaluation, and result reporting are indispensable and contribute to a correct diagnosis. Particularly, the appropriate interpretation of clonality assays requires study, learning, and training on the job. This can be facilitated by participation in EuroClonality educational workshops, or by submitting difficult cases via the EuroClonality website to get online support. In this paper, we show that participation in the EuroClonality EQA schemes significantly contributes to improving the diagnostic interpretation.

The overall performance scores for both IG and TR analysis were high, with more than 90% of successful participants each year. The individual participants had a significantly higher score when participating in more EQA rounds, although there was no obvious overall improvement between 2014 and 2018. There was no observed difference in performance based on the sample type or final clinical diagnosis. The used methodology could not be linked to the improvement for a specific laboratory or sample type.

The final molecular interpretations such as “clonal” or “polyclonal” were in general scored well. Also, truly challenging cases were included such as clonal cases with polyclonal background, cases with bi-allelic rearrangements, bi-clonal cases, and cases with multiple clonal (IGK or TRB) rearrangements that still belong to one clone. In our schemes, thus far, we have evaluated the more detailed molecular interpretations such as bi-allelic or bi-clonal, but no points were deducted when the more detailed interpretation was incorrect. In the next EQA schemes, we intend to include the more detailed interpretations in the evaluation. Based on the previous EQA schemes, we then expect a lower performance for clonal samples. The scoring of oligoclonality clearly was difficult. Oligoclonality is defined as the reproducible detection of three or more clones. As for the interpretation of clonality (including bi-allelic or bi-clonal cases), this requires the appropriate interpretation of the individual tube results as well as understanding the IG and TR loci and PCR design. Due to the non-quantitative PCR nature and potentially preferential amplification of some rearrangements, the identification of true clonal rearrangements compared to minor peaks in an irregular polyclonal background is especially difficult for oligoclonal cases. Because true oligoclonal cases are scarce, the experience with these cases is limited. Only three oligoclonal cases could be included in the paper-based EQA; one for IG and TR in 2015 and one TR case in 2016. Of the total 164 molecular interpretations (Table 3), EuroClonality laboratories correctly assigned the cases as oligoclonal in 24/73 (32.9%) tests, compared to 10/91 (11.0%) for non-EuroClonality laboratories. Of the 56 laboratories interpreting the first oligoclonal TR case in 2015, 39 participated again in 2016. Twenty-eight of 39 participants incorrectly denoted the oligoclonal case in 2015, of which 21 made the same mistake and 7 laboratories improved in 2016. The other 7 participants had a correct outcome in both 2015 and 2016, while 4 laboratories were correct in 2015, but incorrect in 2016.

A better performance (although not significant) was observed for paper-based tests in which only the result interpretation was evaluated, versus the wet cases in which the technological approach, performance of the test, and the interpretation of the results were evaluated. While paper cases were evaluated adequately during a laboratory’s first or second EQA participation, wet cases were more error-prone (Fig. 2, panel B). This is not surprising, given that wet sample analysis includes extra (pre-)analytical processes potentially impacting the results, compared to solely interpreting GS results according to guidelines. However, paper-based cases may also be perceived as difficult by the laboratories, as cases with complex rearrangement patterns were included.

Both EuroClonality and non-EuroClonality laboratories improved their performance upon repeated participation, which is in line with results for biomarker analysis in colorectal cancer [28]. The overall scores were significantly better for EuroClonality laboratories (Fig. 2, panel C). The EuroClonality-affiliated participants may have benefited from the annual meeting and the provided feedback in the group discussions. Feedback has been shown to be an important parameter in learning [32]. However, the better performance should be interpreted with caution, as the majority of the EuroClonality-affiliated laboratories also participated more frequently to the EQA schemes, and repeated participation significantly improved performances. In addition, the difference between EuroClonality and non-EuroClonality laboratories is smaller for individual scheme years. We expected a larger score difference in a single scheme, given the longstanding experience of the EuroClonality participants, who were involved in the design and testing phase of the BIOMED-2/EuroClonality assays and in preparation of the EuroClonality guidelines. The question remains how non-EuroClonality laboratories educated themselves. Most likely, training on the job within an expert environment and/or attending dedicated trainings resulted in translation of theoretical knowledge into diagnostic practice and competence building. In addition, the feedback given in the extended EQA report may also have contributed to learning and good performance [33, 34]. In the end, it remains the responsibility of the participants to implement the necessary corrective actions to improve performance.

This EQA scheme with five wet cases and five paper-based cases in each round allowed us to evaluate the successful performance over two EQA rounds, as it was previously estimated that at least 10 samples are needed to allow a reliable performance estimate [25, 26]. Accredited laboratories have to demonstrate their performance, but not all laboratories participated to all EQA rounds, as the frequency of participation is not specified by ISO15189 [15] or equivalent national accreditation standards. Recent recommendations from the Belgian Molecular Diagnostics working group now state that laboratories should perform a risk analysis to determine their ideal participation frequency [35].

The several international clonality EQA providers each (i) evaluate a different number of IG/ TR targets, (ii) distribute various numbers of samples per annum, and (iii) apply different criteria for successful participation (Supplemental Table 1). Although not all providers include paper-based cases, the majority assesses the laboratory’s interpretation of the rearrangement patterns according to the guidelines. Namely, EQA providers should assess the complete analysis process, from pre-analytical to post-analytical phase [25]. As the participants received pre-extracted DNA samples, the DNA extraction and preparation steps are not evaluated in the EQA scheme, and could impose additional difficulties in a routine setting. The cut-off of 80% for successful performance and 90% after two participations in these EQAs was based on the requirements for EQA programs, which recommend a cut-off of 90% assessed on a total of 10 samples [25]. Similar to the harmonization efforts in molecular oncology, increased harmonization between providers is advisable for clonality analysis to define a uniform scope, scoring system, criteria for successful participation, and actions following unsatisfactory performance in Europe [36].

In summary, we observed a high performance for IG and TR analysis, which increased when participating to more EQA rounds. There was a higher performance for paper-based cases compared to wet cases and for EuroClonality compared to non-EuroClonality laboratories. There was no difference related to the EQA scheme year, sample origin, or clinical diagnosis. The observed difficulties in interpreting oligoclonal cases highlight the need for continued education via meetings and EQA schemes.