National survey on internal quality control for tumour markers in clinical laboratories in China

Introduction This survey was initiated to obtain knowledge on the current situation of internal quality control (IQC) practice for tumour markers (TMs) in China. Additionally, we tried to acquire the most appropriate quality specifications. Materials and methods This survey was a current status survey. The IQC information had been collected via online questionnaires. All of 1821 clinical laboratories which participated in the 2016 TMs external quality assessment (EQA) programme had been enrolled. The imprecision evaluation criteria were the minimal, desirable, and optimal allowable imprecisions based on biological variations, and 1/3 total allowable error (TEa) and 1/4 TEa. Results A total of 1628 laboratories answered the questionnaires (89%). The coefficients of variation (CVs) of the IQC of participant laboratories varied greatly from 1% (5th percentile) to 13% (95th percentile). More than 82% (82 - 91%) of participant laboratories two types of CVs met 1/3 TEa except for CA 19-9. The percentiles of current CVs were smaller than cumulative CVs. A number of 1240 laboratories (76%) reported their principles and systems used. The electrochemiluminescence was the most used principle (45%) and had the smallest CVs. Conclusions The performance of laboratories for TMs IQC has yet to be improved. On the basis of the obtained results, 1/3 TEa would be realistic and attainable quality specification for TMs IQC for clinical laboratories in China.


Introduction
Although the diagnosis of cancer is mostly confirmed by biopsy which has been considered as "gold standard" for a long time, tumour markers (TMs) have an important role in staging and treatment of the cancer (1). Internal quality control (IQC) plays a significant role in the routine practice of clinical laboratories. The central role of IQC is to detect clinically important errors and evaluate repeatability in the analytical process. Only when the imprecisions of the measurement system in the laboratory are small enough, the staff might have the opportunity to get satisfactory and reliable results. Clinical laboratories evaluate the im-precisions of their own measurement systems by monitoring monthly (current) and long-time (cumulative) coefficients of variation (CVs) of IQC data, and the CVs are compared with different quality specifications (allowable imprecision criteria). The results of comparison could give the laboratories directions and suggestions to make their performances better. There are several standards which could be used to evaluate the CVs of IQC, such as the specifications based on biological variations including the minimal, desirable, and optimal allowable imprecisions, and 1/3 total allowable error (TEa) and 1/4 TEa. As the evaluation standards of the external quality assessment (EQA) in China were often set as TEa, the 1/3 TEa and 1/4 TEa could be calculated easily for each analyte whose TEa has been known before. Clinical laboratories in China have always used 1/3 TEa and 1/4 TEa to evaluate the imprecisions of their measurement systems (2). Although the 1/3 TEa and 1/4 TEa were convenient and easy to use, consideration of combining allowable imprecision specifications based on biological variations and different clinical requirements might be more suitable for all kinds of clinical laboratories regardless of size and conditions, which had gained a consensus recommendation among experts and clinical laboratories staff in recent years (3). From each TMs EQA program participating laboratory in China two types of CVs, the control rules, methods, instruments, reagents, calibrators and averages of 6 TMs (alpha-fetoprotein (AFP), carcinoembryonic antigen (CEA), total prostate specific antigen (PSA), cancer antigen 125 (CA125), cancer antigen 15-3 (CA15-3) and cancer antigen 19-9 (CA19-9)) IQC materials were collected. Then all the acquired CVs of IQC were analysed per 5 imprecision specifications (minimal, desirable, and optimal allowable imprecision based on biological variations, and 1/3 TEa and 1/4 TEa). After that we could get an overview of measurement imprecision for these 6 TMs in clinical laboratories in China. So far, IQC was one of the best ways to evaluate the imprecision of routine laboratory work in simple, convenient and effective way. There are some different measurement systems used to test TMs in China. We wanted to evaluate and compare their performances, and gain insight on which one is used the most and which one had the smallest imprecision.

Materials
Clinical laboratories participating in the 2016 TMs EQA programme, organized by the National Center for Clinical Laboratories (NCCL) which had been the official EQA programs provider in China for more than 20 years, received survey questionnaires. A total of 1821 laboratories received the questionnaire and 1628 answered it (89%). The TMs included were: AFP, CEA, PSA, CA125, CA15-3 and CA19-9.

Methods
Our study was a current status survey.

Imprecision criteria
Based on the performance evaluation requirements provided by Clinical Laboratory Improvement Amendment (CLIA' 88), our institute set these criteria as the TEa for these 6 TMs. The 1/3 TEa and 1/4 TEa could be calculated and used to evaluate the imprecisions of the IQC of the participant laboratories in China. There were three imprecision levels which were calculated from biological variations and used to evaluate the performance of IQC, including: 1) minimum performance defined by CV A < 0.75C V I (CV A is analytical precision and CV I is the within-subject biological variation); 2) desirable performance defined by CV A < 0.50 CV I ; 3) optimum performance defined by CV A < 0.25 CV I . The data we used referred to the biologic variation list provided by Ricos et al. (4).

Statistical analysis
Data were analysed using SPSS 13.0 (SPSS Inc, Chicago, IL, USA) and Clinet-EQA evaluation system V1.0 (Clinet Corp, Beijing, China), designed by NCCL and used in the national EQA program (see http://www.clinet.com.cn/shop/shop). The distributions of CVs of each analyte were tested by Kolmogorov-Smirnov test for normality. The related statistical parameters of CVs, including median, the 5th, 25th, 75th, and 95th percentile were calculated. The percentages of laboratories (total and divided into different subgroups by measurement systems) that met quality requirements of imprecision were calculated.

Results
The questionnaire was answered by 1628 laboratories (1628/1821, 89%). Laboratories which replied the survey submitted the related information and data for more than one TM. Table 1 shows the definitions of survey items. Table 2 shows the quality specifications of TMs based on CLIA'88 and biological variations, while Table 3 shows the numbers of participant laboratories, medians, other percen-  tiles of two types of CVs, and the percentages of laboratories meeting quality requirements. Table 4 shows the principles and systems used for testing TMs; Table 5 and 6 show the percentages of two types of CVs of different principles and systems meeting different imprecision criteria. These two types of CVs were shown abnormal distributions accessed by normality test.
In Table 3 current CVs are listed and 5 out of 6 analytes (except CA19-9) showed relatively smaller CVs compared with CA19-9. The percentages of laboratories whose CVs were smaller than 1/3 TEa specification for these TMs were all above 87% (from 87% for PSA to 91% for CA125), while the percentages varied markedly (from 3% for CA15-3 to 74% for CA125) when the optimal allowable imprecision specification was used. Although current CVs of CA19-9 were larger than other TMs, the percentage of laboratories whose CVs were smaller than 1/3 TEa specification was close to 80% (974/1247). The percentages of laboratories whose CVs met imprecision criteria based on biological variations for CA15-3 were significantly smaller than other TMs. For the cumulative CVs, as the current ones, 5 TMs got relatively smaller CVs compared with CA19-9. The percentiles of current CVs of laboratories were lower than cumulative CVs.

Discussion
The results of our survey suggest that there were remarkable variations for the TMs IQC in China, including manufacturer, principle and imprecision. Harmonized control rules with defined ranges of imprecisions were not defined in China. Some studies only kept their focus on the performances of different test systems (5). Compared with Bertsch et al. the range of CVs for CA19-9 IQC was much wider in our study. The performances should be improved later (5). Consequently, the lack of standards might lead to many problems in daily monitoring of imprecision. The wide ranges of current and cumulative CVs of TM IQC materials reported by the participant laboratories, which varied dramatically (from less than 1% to more than 50%) also shocked us. TM -tumour marker. Group 1 -acridine-ester direct chemiluminescence. Group 2 -microparticle chemiluminescence. Group 3luminol/isoluminol chemiluminescence. Group 4 -flow fluorescence immunoassay. Group 5 -electrochemiluminescence. Group 6 -enzyme immunochemical luminescence. TEa -total allowable error. AFP -alpha-fetoprotein. CEA -carcinoembryonic antigen. PSA -total prostate specific antigen. CA125 -cancer antigen 125. CA15-3 -cancer antigen 15-3. CA19-9 -cancer antigen 19-9. TM -tumour marker. Group 1 -acridine-ester direct chemiluminescence. Group 2 -microparticle chemiluminescence. Group 3luminol/isoluminol chemiluminescence. Group 4 -flow fluorescence immunoassay. Group 5 -electrochemiluminescence. Group 6 -enzyme immunochemical luminescence. TEa -total allowable error. AFP -alpha-fetoprotein. CEA -carcinoembryonic antigen. PSA -total prostate specific antigen. CA125 -cancer antigen 125. CA15-3 -cancer antigen 15-3. CA19-9 -cancer antigen 19-9. Cancer antigen 19-9 had the largest CVs of all TMs, which presented that Chinese laboratories should pay more attention to this analyte. The cumulative CVs were slightly larger than current CVs, which might reveal that the long-time repeatability was not as good as short-term one. We might provide some suggestions such as: 1) about 80% Chinese clinical laboratories with better medical resource could meet the 1/3 TEa criteria which could be considered as evaluation standard for IQC program; 2) the cumulative CVs were larger than current CVs, which means that more attention should be paid to long-time repeatability. The ideal situation should be, the longer a lot of IQC material tested, the smaller CVs should be. That was because with the increasing of test times, the results and variation should be more and more stable, if the test system had been stable all the time. The long-time stability of measurement system should be improved in China.
Cancer antigen 15-3 had the strictest specification based on biological variation, which made the percentages of laboratories meeting criteria much lower than other TMs. So we should know that 1/3 TEa should be used carefully because it was much bigger than the specifications based on biological variation. We might pay more effort on finding a better and more reasonable IQC evaluation criterion which could better match the performance of test system and biological variations in the future.
The ECL method had the highest CVs of TMs IQC meeting the evaluation specifications, which might be one of the reasons why most laboratories chose ECL method. A study also verified that the ECL method and Roche test system could achieve a good performance in TMs measurement (6). Compared with other analytes in Chinese Proficiency testing (PT) panel, creating national evaluation specifications for imprecision of TMs was needed to improve the quality of measurement (7,8).
Clinical laboratories in this study were located in different areas of China and could be considered as representatives of Chinese clinical laboratories with better medical resources and more concern on their performance than those who did not participate to TMs EQA program or return the questionnaires. Most of the clinical laboratories participating in these kinds of surveys were mostly firstclass hospitals in China (9,10).
There were still some limitations in our study. Human epidermal growth factor receptor 2 (HER2) was a TM which was newer than these 6 TMs and it had been in focus in recent years. There is another way to perform IQC for HER2 and we should keep up with the time (11). Another study reported that lyophilized QC materials for TMs were insufficiently stable for use in quality control among clinical laboratories (12). This might result in cumulative CVs bigger than current ones, which might be studied in the further studies. In Lent et al. study, they made a conclusion that treatment errors in association with PSA determinations could therefore be uniformly and plausibly assessed using objective criteria and could thus be avoided (13). IQC was an effective way to decrease errors occurrence. We might combine IQC to evaluate treatment errors occurrence rate to study the relationship between them.
In conclusion, the performance of laboratories for TMs IQC has yet to be improved. On the basis of the obtained results, 1/3 TEa would be realistic and attainable quality specification for TMs IQC for clinical laboratories in China.