CD117 (c-KIT) is a tyrosine kinase receptor protein, mutations in which are important in the tumourigenesis of gastrointestinal stromal tumour (GIST). Immunohistochemical detection of CD117 is the primary identifying feature in its diagnosis.
Introduction
UK NEQAS ICC & ISH conducts an EQA programme for CD117 expression in GIST. Between May-2020 and May-2025, 20 runs of this EQA were undertaken. The archived data was analysed to look for evidence regarding the quality of testing and associations between primary antibodies and stain quality.
Discussion
Total submissions were 2,656 (mean per run = 132.8; range = 125–140). The number of submissions awarded a quality score indicating at least acceptable stain quality was 2,642 (99.5%). The total number of submissions that failed was 14 (mean proportion per run = 0.5%; range = 0.0%–2.4%); those obtaining a borderline score totalled 158 (mean proportion = 5.9%; range = 0.7%–10.9%); those obtaining an acceptable score totalled 394 (mean proportion = 14.8%; range = 4.3–24.4%); and those obtaining a good/excellent score totalled 2,090 (mean proportion = 78.7%; range = 57.7%–94.2%). There was a strong trend for the proportion of submissions obtaining a good/excellent quality score to increase over time. Four primary antibodies were used: Clone 9.7 (102 submissions, 3.9%), YR145 (263 submissions, 10.0%), EP10 supplied by Leica Biosystems (EP10(LB), 415 submissions, 15.8%), EP10 supplied by Roche Diagnostics (EP10(RD), 505 submissions, 19.2%), and a polyclonal antiserum supplied by Agilent Dako (1,263 submissions, 48.1%). A total of 76 submissions (2.9%) that used other primaries or suppliers were excluded from analysis.
Conclusion
Demonstration of CD117 overexpression in GIST was undertaken to a very high quality standard in the majority of laboratories. For users of Dako Agilent supplied automation the highest quality staining was produced by using the polyclonal antiserum from the same supplier. Among users of Leica Biosystems and Roche Diagnostics automation the best staining was produced by using primary sourced from the same supplier that provided their automation. It is recommended that Clone 9.7 (Roche Diagnostics) should not be used as the results it produced were statistically significantly inferior to those of all other antibodies.
CD117EQAGISTimmunohistochemistryqualityexternal quality assessmentThe author(s) declared that financial support was not received for this work and/or its publication.Introduction
CD117 was first reported to be a sensitive marker of gastrointestinal stromal tumours (GISTs) by Miettinen et al at the end of the last century [1–3]. It continues to this day to be used as the primary marker in the diagnosis of this important mesenchymal tumour of the gastrointestinal tract [4].
GISTs arise predominantly from the interstitial cells of Cajal situated in the muscle layers of the bowel wall, with most presenting in the stomach or small intestine with either spindle cell or epithelioid morphology [5–9].
The KIT gene encodes CD117 (c-KIT) protein, which is a tyrosine kinase receptor. Mutations in the KIT gene are drivers of GISTs and also explain why they so commonly over-express the protein [10].
The introduction of treatments using tyrosine kinase inhibitors such as imatinib and sunitinib revolutionised the prognosis for patients with advanced unresectable or metastatic disease, it is therefore imperative that GISTs are correctly identified at primary diagnosis [11–13].
Despite its obvious importance, there are very few papers reporting on CD117 stain quality in the demonstration of GIST. One which is relevant was a report from 2006 [14]. In this study conducted by the College of American Pathologists they found that most laboratories that participated produced good quality staining. In their paper published in 2019 the NordiQC EQA organisation presented an analysis of the results from their CD117 quality assessments [15].
The UK National External Quality Assessment Scheme for Immunocytochemistry and In-Situ-Hybridisation (UK NEQAS ICC & ISH) [16] conducts quality assessment of the testing for GIST using CD117 immunohistochemistry in clinical laboratories in the UK and overseas. Its archives were surveyed, looking at the results from the last 5-years with the aim of producing evidence on the comparative performance of the immunohistochemical methods that are in clinical use.
Materials and methods
The Scheme’s Alimentary Tract Pathology EQA Programme which looks at the immunohistochemical staining of markers for CD117 in GIST was examined. Assessment runs were conducted regularly between May 2020 and May 2025 (N = 20). Data on performance and immunohistochemical staining methodology were retrieved from the Scheme’s archives for each participant at each run.
Participant characteristics
Participants in the EQA Programme were predominately healthcare-based cellular pathology departments undertaking immunohistochemical assessment of CD117 in the clinical setting. They were based in both the UK and overseas, with an almost equal split in frequency.
EQA run design
Each participant was provided with a slide bearing unstained sections taken from formalin-fixed paraffin-embedded (FFPE) tissue samples. The samples comprised:
Appendix, useful for assessing staining interstitial cells of Cajal and mast-cells both of which were expected to show specific staining, and for assessing non-specific staining elsewhere.
A GIST, previously tested to ensure CD117 positivity.
A relevant alternative tumour such as gastric adenocarcinoma, leiomyosarcoma or desmoid tumour, used to detected non-specific staining.
After undertaking immunohistochemical staining for CD117 using their in-house standard operating procedures, participants returned their stained slide to UK NEQAS ICC & ISH for collation and central visual assessment (see below for further details on the assessment process).
Quality assessment metrics
The Scheme uses a scoring method at assessment that is based on four expert assessors evaluation. The four assessors score each participant’s submitted stained slide concurrently but independently using a multi-header microscope. Each assessor awards a score within a range of 1–5, with 1 being of no utility and 5 being excellent quality. These scores are summed for all four assessors to give a score in the range 4–20 (see Table 1 for further details).
The Scheme’s qualitative scoring system.
Overall Quality Score
Individual Assessor Score
Interpretation
4–8
1
Unreadable. Staining has no diagnostic utility. Improvement essential
2
Substantially sub-optimal preparation. Staining has no diagnostic utility. Improvement essential
12
3
Sub-optimal but readable preparation. The staining quality would allow the correct diagnosis to be achieved but requires substantial improvement
16–20
4
Good preparation. Staining appropriate for the target and of good quality. Only minor improvement(s) are possible
5
Excellent preparation. Staining appropriate for the target and of excellent quality. No improvement required
Four assessors scored independently based on their visual assessment of the preparation’s stain quality. Assessor’s scores were only allowed to vary by one mark i.e., scores of 3 and 4 might be awarded by individual assessors scoring the same preparation, but not scores of 3 and 5; when scores between assessors did show such variations the quality of staining was discussed and a consensus was agreed. If an assessor wished to fail a participant’s submission i.e., award a score of 1 or 2, then all assessors must have agreed and awarded scores in the same range.
Samples were assessed against expected expression levels, which were established beforehand using preparations stained by the commercial suppliers of commonly used primary antibodies and in-house stained slides.
Sections, which had been cut at the same time as those distributed to participants were stained on the slide return deadline date to ensure the stability of expression.
To look for biological variation due to tumour heterogeneity sections taken at levels through the block were stained and assessed.
Data analysis
Descriptive statistics were produced looking at trends in primary antibody usage over time.
The association of stain quality with key following elements of the immunohistochemical staining process were examined:
Primary antibody (clone and supplier)
Antigen retrieval (type, duration and supplier)
Detection system (type and supplier)
Automated platform (type and supplier)
The t-test (Mann-Whitney for non-parametric data) was used to assess level of significance (P < 0.05).
Data were collated and analysed using Excel (Office 365, Microsoft), which was also used for the preparation of charts and tables. Statistical analyses and box and whiskers chart production used the following software packages: SPSS (Version 29.0.2.0 (20), IBM) and Prism (Version 10.3.0, GraphPad).
Results
In the period between May 2020 and May 2025 the Scheme conducted 20 assessment runs. The total number of submissions made was 2,656 (mean number of submissions per run = 132.8; range = 125–140). During the 5-year period of the study the number of participants rose from 125 to 140 (12.0% increase).
Overall quality score results
The total number of submissions which were awarded a quality score indicative of at least acceptable stain quality (score ≥12/20, a “pass”) was 2,642 (99.5%).
The total number of submissions which failed assessment was 14 (score = 4–8; mean proportion per run = 0.5%; range = 0.0–2.4%), those obtaining a Borderline score totalled 158 (score = 12; mean proportion = 5.9%; range = 0.7–10.9%), for those obtaining an Acceptable the total was 394 (score = 13–15; mean proportion = 14.8%; range = 4.3–24.4%), and for those obtaining a Good or Excellent score the total was 2,090 (score = 16–20; mean proportion = 78.7%; range = 57.7–94.2%).
There was a strong trend for the proportion of submissions which obtained a Good or Excellent quality score to increase over time. See Table 2; Figure 1 which show the data in detail.
Descriptive statistics.
Run date
Fail (4–8)
Borderline (12)
Acceptable (13–15)
Good/Excellent (16–20)
Run total
N
%
N
%
N
%
N
%
N
% (*)
May-2020
3
2.4
12
9.6
27
21.6
83
66.4
125
4.7
Aug-2020
0
0.0
14
10.9
22
17.2
92
71.9
128
4.8
Nov-2020
1
0.8
13
10.2
31
24.4
82
64.6
127
4.8
Feb-2021
2
1.5
9
6.9
44
33.8
75
57.7
130
4.9
May-2021
1
0.8
10
7.8
26
20.3
91
71.1
128
4.8
Aug-2021
2
1.6
10
7.8
33
25.6
84
65.1
129
4.9
Nov-2021
1
0.8
11
8.4
20
15.3
99
75.6
131
4.9
Feb-2022
0
0.0
5
3.8
13
9.8
114
86.4
132
5.0
May-2022
0
0.0
10
7.8
17
13.3
101
78.9
128
4.8
Aug-2022
1
0.8
10
7.8
27
21.1
90
70.3
128
4.8
Nov-2022
0
0.0
8
6.3
10
7.8
110
85.9
128
4.8
Feb-2023
1
0.8
4
3.1
21
16.2
104
80.0
130
4.9
May-2023
1
0.7
4
2.9
17
12.3
116
84.1
138
5.2
Aug-2023
0
0.0
8
5.8
14
10.1
117
84.2
139
5.2
Nov-2023
0
0.0
2
1.4
6
4.3
131
94.2
139
5.2
Feb-2024
0
0.0
4
2.9
11
7.9
124
89.2
139
5.2
May-2024
0
0.0
4
2.9
14
10.1
120
87.0
138
5.2
Sep-2024
1
0.7
12
8.6
16
11.5
110
79.1
139
5.2
Jan-2025
0
0.0
1
0.7
13
9.3
126
90.0
140
5.3
May-2025
0
0.0
7
5.0
12
8.6
121
86.4
140
5.3
Total
14
158
394
2090
2656
Mean
0.5
5.9
14.8
78.7
100.0
Showing for each Run the number (N) and proportion (%) of submissions assessed to be in each quality score category, together with the total number of submissions at that Run. The figure shown in the Run Totals column indicated by the ‘*’ symbol is the proportion of the total submissions. See also Figure 1.
Descriptive statistics. The data shown in Table 2 are represented graphically here. For clarity data labels are presented for the overall proportions of submissions receiving a passing score only, for more details see Table 2. The blue-dotted line is the linear trendline for the proportion of submissions receiving a score indicative of a good or excellent stain quality (score 16–20).
Line graph displaying proportions of assessment submissions in different grade categories over time from May 2020 to May 2025. Overall pass rates shown in black remain near 100 percent throughout. Good/Excellent scores in green show an upward trend, illustrated with a dotted linear trendline. Acceptable scores in gray fluctuate between 10 and 30 percent. Borderline and Fail categories, in orange and red respectively, remain consistently low. The x-axis lists assessment dates, and the y-axis indicates the proportion of submissions in percentage.Primary antibodies
Details on the primary antibodies used were returned for 2,624 submissions (98.8%).
Four rabbit monoclonal and two rabbit polyclonal reagents were used to detect CD117. Three of the four monoclonals (Clone 9.7, DBRRM1.17 and YR145) were each obtained from a single different commercial supplier respectively, while EP10 was sourced from five different suppliers. The two polyclonal antisera were each supplied by two different suppliers. See Table 3 for details and references.
Primary antibodies: descriptive statistics.
Primary Antibody and Supplier
Product Code
Type
Format
Count (N)
Proportion (%)
Clone 9.7 [17]
102
3.9
Roche diagnostics
790–2951
Rb Mab
RTU
102
3.9
DBRRM1.17 [18]
5
0.2
Diagnostic biosystems
RMAB119
Rb Mab
RTU
5
0.2
EP10 [19–21]
982
37.4
Biocare Medical
PME-296AA
Rb Mab
Concentrate
18
0.7
CliniSciences
MAD-000644QD
Rb Mab
RTU
1
0.0
Epitomics
AC-0029
Rb Mab
Concentrate
7
0.3
Roche diagnostics
790–7061
Rb Mab
RTU
505
19.2
Leica biosystems
CD117-032-L-CE
Rb Mab
Concentrate
36
1.4
Leica biosystems
PA0007
Rb Mab
RTU
415
15.8
Polyclonal [22, 23]
1,272
48.5
Agilent dako
A4502
Rb pab
Concentrate
1,263
48.1
Generon
PDR045
Rb pab
Concentrate
9
0.3
YR145 [24, 25]
263
10.0
Cell Marque
117R-18
Rb Mab
Concentrate
263
10.0
Total
2624
100.0
Details of the primary antibody reagents used. Showing both the aggregated data for each primary antibody reagent and the breakdown by supplier and format. For 32 submissions the participant did not supply methodological data. Rb Mab = rabbit monoclonal; Rb Pab = rabbit polyclonal; RTU = ready-to-use. Bold values indicate summations for that category.
The primary antibody reagents were supplied as either prediluted ‘ready-to-use’ (RTU) or concentrated formulations. Overall the total number of submissions that used RTU reagents was 1,028 (39.2%), while those using a diluted solution taken from a concentrate totalled 1,596 (60.8%). Table 3 shows this and additional descriptive data for the primary antibodies.
In order to present the significant results clearly, individual antibody reagent/supplier combinations represented at a frequency of <100 submissions (<3.9%) were excluded from further discussion. The number of submissions for the six reagent/supplier combinations that were excluded had a range of 1–36 (mean: 12.6), thus even the most prevalent of them was used by less than two participants at any given run. Any conclusions derived from analysis of their data would have been entirely anecdotal.
Five primary antibody reagents will be discussed: Clone 9.7 (102 submissions, 3.9%), YR145 (263 submissions, 10.0%), EP10 supplied by Leica Biosystems (EP10(LB), 415 submissions, 15.8%), EP10 supplied by Roche Diagnostics (EP10(RD), 505 submissions, 19.2%), and a polyclonal antiserum supplied by Agilent Dako (1,263 submissions, 48.1%). In total 76 submissions (2.9%) were excluded from the subsequent analyses.
Primary antibody performance
The analysis of quality score data indicated that the demonstration of CD117 in the UK NEQAS provided materials was good. For the whole population the mean quality score was 16.3 (95% CIs: 16.1–16.4).
Primary antibody usage
the use of the primary antibody reagents over the 5-year period of the study was examined, with consistent trends been seen. The proportional use of the polyclonal antiserum declined from 58.7% in 2020 to 42.4% in 2025. There was also a decline in the use of the 9.7 clone from 12.4% to 0.8%, and the YR145 clone from 10.7% to 8.8%. In contrast the EP10 clone showed an increase in the proportion of users from 18.2% (3.3% for the Roche Diagnostics supplied reagent and 14.9% for the Leica Biosystems supplied reagent) to 48.0% (30.4% for EP(RD) and 17.6% for EP(LB)), such that the EP10 clone became the most frequently used primary reagent at any given assessment during the course of 2025, overtaking the polyclonal antiserum which had consistently held that position previously. See Figure 2, which illustrates these data.
Trends in proportional use of primary antibodies over time. The change proportion of submissions using each of the primary antibody reagents is shown per year.
Line graph illustrating the proportions of five primary antibodies used from May 2020 to May 2025. Polyclonal starts highest at 58.7 percent but decreases to 42.4 percent. EP10 (RD) increases steadily from 3.3 percent to 30.4 percent. EP10 (LB) remains relatively stable near 15 percent, YR145 stays near 9 percent, and Clone 9.7 declines from 10.7 percent to 0.8 percent.Factors associated with fails at assessment
Overall, 12 submissions (12/2548, 0.5%) were assessed as fails (quality score = 8/20). When dealing with such a small dataset it is not possible to drawn any firm conclusions about factors associated with assay failure. However, it is worthwhile to note that the assessors assigned only two indications to failed submissions, the first being very weak staining of the GIST sample and normal elements expected to stain i.e., interstitial cells of Cajal and mast cells which was noted in seven submissions (7/12, 58.3%), and the second was excessive non-specific and/or inappropriate staining particularly in the non-GIST tumour and the appendix samples, which was noted for five submissions (5/12, 51.7%).
The polyclonal reagent was used in eight of the failed submissions (8/12, 75.0%), but the preponderance is not surprising given that the polyclonal antiserum was used in just under half of all submissions made (1,263/2548, 49.6%) and was present in the data set more than twice as frequently as the next most “popular” antibody (EP10(RD), 505/2548, 19.8%). See also Table 4.
Factors present in submissions that failed at assessment.
There were clear difference in the degree to which primary antibody and automation supplier were associated. For Roche Diagnostics there was an absolute association for both Clone 9.7 and EP10(RD), and for Leica Biosystems it was extremely high (94.9%) for the EP10(LB) reagent. In contrast the polyclonal antiserum supplied by Agilent Dako was used on machines manufactured by the same supplier with a frequency of only 26.7%. Indeed it was used more frequently on both Leica Biosystems (27.6%) and Roche Diagnostics supplied automation (45.8%).
The YR145 clone (supplied by Cell Marque) was used primarily in conjunction with Roche Diagnostics supplied automation (82.5%). See also Table 5.
Association of Primary antibody supplier and Automation supplier.
Primary antibody clone
Primary antibody supplier
Agilent Dako
Leica biosystems
Roche diagnostics
Totals
N
%
N
%
N
%
N
Clone 9.7
Roche diagnostics
0
0.0
0
0.0
102
100.0
102
EP10(RD)
Roche diagnostics
0
0.0
0
0.0
505
100.0
505
EP10(LB)
Leica biosystems
20
4.8
393
94.9
1
0.2
414
Polyclonal
Agilent dako
337
26.7
348
27.6
578
45.8
1,263
YR145
Cell Marque
27
10.3
19
7.2
217
82.5
263
Totals
384
15.1
760
29.8
1,403
55.1
2547
Correlating the supplier of the primary antibody with the supplier of the automation it was used on. The grey-shaded boxes highlight the matched combinations. N = count of submissions, % = proportion of submission. Data for one submission has been omitted as the automation type was not submitted.
In light of the use of the polyclonal antiserum across all major suppliers’ platforms, a situation which differed markedly from that seen in the use of both EP10(LB) and EP10(RD), an analysis was undertaken to look for an association of performance with individual platform supplier. The mean quality score was highest when the polyclonal primary was used on Agilent Dako machines (16.6, 95% CIs: 16.4–16.8), when used on the Leica Biosystem machines it was 16.1 (95% CIs: 16.0–16.3), and on Roche Diagnostic supplied machines it was 16.2 (95% CIs: 16.1–16.4). See Figure 3.
Box and whiskers plot illustrating quality score distributions for polyclonal primary antibody used on different automated platforms. Box heights indicate the 25% and 75% interquartile range, the whiskers show the 5% and 95% percentiles. The lines within boxes show the median and the crosses the mean quality scores.
Boxplot graphic compares quality scores for three automation suppliers: Agilent Dako, Leica Biosystems, and Roche Diagnostics. Each box shows median, quartiles, and outliers for quality scores ranging from zero to twenty.
The difference in quality score distributions was statistically significant when comparing Agilent Dako and Leica Biosystems (P < 0.0001) and Agilent Dako and Roche Diagnostics (P = 0.0013). It was not different when Leica Biosystems and Roche Diagnostics machines were compared (P = 0.216).
A follow-on analysis looked at the results for submissions which used primary antibody reagent and the automation matched for supplier. It found that EP10(LB) on Leica Biosystems machines and the polyclonal antiserum on Agilent Dako machines were both associated with superior quality staining results compared to those produced by EP10(RD) on Roche Diagnostics machines. But they were not statistically different from each other. See Figure 4; Table 8.
Forest plot illustrating quality score distributions for primary antibodies and matched automated platforms. The diamond shaped boxes indicate the mean quality scores; the whiskers show the 95% confidence intervals. AD = Agilent Dako, LB = Leica Biosystems, RD = Roche Diagnostics.
Horizontal error bar chart compares quality scores of three primary antibodies, showing means and confidence intervals. Labels on the y axis are “Polyclonal on AD,” “EP10 (LB) on LB,” and “EP10 (RD) on RD.” Quality score ranges from sixteen to seventeen.Discussion
The recent data produced by UK NEQAS ICC & ISH in its CD117 in GIST EQA Programme was analysed to look for evidence on the performance of the primary antibody reagents that are used and their effect on the quality of staining produced. The main aim was to produce robust evidence-based information that could be used by those undertaking clinical testing to ensure they are achieving the highest quality results possible, and in so doing assure diagnostic outcomes.
The data available to us was drawn from immunohistochemistry methodology records that accompanied the stained slides submitted over the course of twenty GIST EQA assessment runs that were conducted over a 5-year period between May 2020 to May 2025. During that time the Scheme assessed the quality of 2,656 submissions and received data on the methodology used for 2,624 of them. Having such large datasets made it possible to find associations between the quality and the method that were highly statistically significant.
The overall proportion of submission receiving at least a borderline acceptable quality score was 99.5% with the lowest at any run being 97.6%, indicating that this marker was demonstrated to an adequate standard in the vast majority of laboratories. However this does not mean there was no room for improvement, as is indeed reflected in the proportion of submissions receiving a score indicating good or excellent staining, which is the standard all laboratories should be aiming to achieve. This was 78.4% on average (range 57.7%–94.2%) i.e., at one point only just over half of submissions demonstrated good or excellent staining.
Our survey has captured a period over which a dynamic change in the quality of staining occurred, as can be clearly appreciated when the data presented in Figure 1 are reviewed. There was a noticeable sustained increase in the proportion of submissions showing good or excellent staining from 65% at the first run in 2020 to 90% in the last in 2025. It is probable that a major proportion of the improvement can be attributed to the concomitant change in the landscape of primary antibody usage that occurred, this will be discussed in more detail below.
Four primary antibody clones and one polyclonal antiserum were reported as being used, and as is usually seen these were supplied by one or other of the three main commercial suppliers in the field, namely Agilent Dako, Leica Biosystems and Roche Diagnostics. There was one notable exception in the YR145 clone that was supplied by Cell Marque. By not including them in the analyses no disrespect is intended to companies which supplied antibodies that were only used on a small number of occasions, the necessity to make evidence-based associations precluded them.
Historically, the polyclonal antiserum (Agilent Dako, A4502) dominated the field, being the first commercially available primary to be widely reported as being effective in the demonstration of CD117 in formalin-fixed paraffin embedded (FFPE) samples. In this study submissions using it represented just less than half of the total. However, it is notable that its proportional use showed a steady decline over the course of the study (from 58.7% to 42.4%), such that the monoclonal primaries when taken in aggregate now hold the larger share of the market.
The rabbit monoclonal antibody EP10, available from a variety of suppliers but notably in our data from Leica Biosystems as an RTU (PA0007) and Roche Diagnostics also as an RTU (790–2951) saw an increase in their use. The Roche Diagnostics supplied reagent hereafter referred to as EP10(RD) saw a ten-fold increase proportional use from 3.3% to 30.4%, such that it became the second-most commonly used primary antibody. For the most-part the increase in use was due to laboratories that had previously used the rabbit polyclonal antiserum changed to using EP10(RD), and this was particularly prevalent amongst laboratories that used Roche Diagnostic automation. Taking EP10(RD) and the Leica Biosystems supplied reagent (EP10(LB)) together, the proportion of EP10 users was larger than that of the polyclonal antiserum at the time the survey period ended (48.% compared to 42.4% respectively).
A third notable change in primary antibody use was the decline in use of the rabbit monoclonal antibody 9.7 (Roche Diagnostics, RTU, 790–2951), which had a 12.4% share of users among participants in 2020 that fell to just 0.8% in 2025. Again this was due principally to participants changing to EP10(RD).
When the performance of the primaries was examined using higher quality score as the indicator of superior performance the performance of the EP10(LB) reagent was 16.9 (95% CIs: 16.7–17.0). No submission that used this primary failed at assessment, which was in contrast to all other primary reagents where there was at least one fail. Its performance was superior to that of all other primary reagents, with the difference being significant at P < 0.05 when compared to Y145 and highly significant (P < 0.0001) when compared to both EP (RD) and the polyclonal antiserum.
Submissions that used Clone 9.7 had the worst performance with a mean quality score of 14.6 (95% CIs: 14.2–15.0). There was a highly significant association between the use of this antibody clone and lower quality scores (P < 0.0001), when its performance was compared to all other reagents.
The Y145, the EP (RD) and the polyclonal antiserum produced results that were not significantly different from one another. See Table 6; Figure 5, which show antibody performance data in detail and Table 7 which shows the statistical significance analysis results.
Primary antibodies: quality score analysis.
Descriptive statistic
Clone 9.7
YR145
EP10 (LB)
EP10 (RD)
Polyclonal
All
Count (N)
102
263
415
505
1,263
2548
25% percentile
13.0
16.0
16.0
16.0
16.0
16.0
Median
15.0
16.0
17.0
16.0
16.0
16.0
75% percentile
16.0
18.0
18.0
17.0
17.0
17.0
Mean
14.6
16.5
16.9
16.3
16.3
16.3
Std. Deviation
2.1
2.1
1.4
1.8
1.9
1.9
Lower 95% CI
14.2
16.2
16.7
16.1
16.2
16.2
Upper 95% CI
15.0
16.7
17.0
16.5
16.4
16.4
Details of the primary antibody performance. Std. = Standard; CI = confidence interval of the mean.
Box and whiskers plot illustrating quality score distributions for primary antibodies. Box heights indicate the 25% and 75% interquartile range, the whiskers show the 5% and 95% percentiles. The lines within boxes show the median and the crosses the mean quality scores.
Boxplot chart comparing quality scores for six primary antibody categories: Clone 9.7, YR145, EP10 (LB), EP10 (RD), Polyclonal, and All. Each colored box shows the score distribution, median, mean (plus sign), and outliers, with quality scores ranging from zero to twenty.
Results of test to compare quality scores.
Primary antibody
Clone 9.7
Y145
EP10 (LB)
EP10 (RD)
Polyclonal
Clone 9.7
<0.0001
<0.0001
<0.0001
<0.0001
Y145
****
0.0250
0.1925
0.0902
EP10 (LB)
****
*
<0.0001
<0.0001
EP10 (RD)
****
NS
****
0.8094
Polyclonal
****
NS
****
NS
The Mann-Whitney test compares the distributions of two unmatched groups at P < 0.05. * = different at P < 0.05; **** = different at P < 0.0001; NS = not significantly different.
As was outlined in the Results section a markedly different pattern was found when the data for primary antibody usage across the machines of the three major automated platform suppliers was examined. It was clear that antibody and platform was very closely coupled for clone 9.7, EP10(LB) and EP10(RD). But this was not the case for the polyclonal reagent from Agilent Dako; the proportion of its users that used it on Agilent Dako sourced automation was actually smaller than the proportions which used it on Leica Biosystems or Roche Diagnostics machines. The data was further examined to see if this affected performance results for the polyclonal antiserum and found that it did. The size of the effect was such that the performance of the polyclonal antiserum when it was used on Agilent Dako automated platforms became equivalent to the best performing antibody, EP10(LB) (no significant difference at P < 0.05) (See Table 8; Figure 3).
Results of test to compare quality scores.
Primary antibody (Supplier)
EP10 (LB) on LB
EP10 (RD) on RD
Polyclonal on AD
EP10 (LB)
<0.0001
0.3425
EP10 (RD)
****
0.0084
Polyclonal
NS
**
Mann-Whitney test to compares the distributions of two unmatched groups at P < 0.05. ** = different at P < 0.005; **** = different at P < 0.0001; NS, not significantly different.
This indicates the importance of analysing the data in a context sensitive way and one that reflects real-world usage.
This work represents an advance in biomedical science because it uses a set of real-world data to produce evidence that can be used to assure laboratory testing quality and improve patient outcomes.
Conclusions
The demonstration of CD117 over-expression in GIST is undertaken to a very high standard of quality in the majority of laboratories that participate in regular scheduled EQA for this important predictive biomarker.
Statistically tested and proven results indicate that for users of Dako Agilent supplied automation the highest quality staining is produced by using the polyclonal antiserum from the same supplier.
Users of Leica Biosystems and Roche Diagnostics automation should use the EP10 rabbit monoclonal primary, and should source that reagent from the same supplier that provides their automation.
It is recommended that Clone 9.7 as supplied by Roche Diagnostics should not be used as the results it produces are statistically significantly inferior to those obtained using all other surveyed primary antibodies.
Summary tableWhat is known about this subject:
CD117 (c-KIT) is an important marker that is used in the diagnosis of gastrointestinal stromal tumours (GISTs).
Mutational analysis of the KIT gene is used to guide therapy choice following a confirmed diagnosis of GIST.
Targeted treatment using imatinib, which is a tyrosine kinase inhibitor, is highly effective in approximately 85% of GISTs.
What this paper adds:
This work presents information to aid in the choice of immunohistochemical methods that optimise testing quality.
The results confirm that the majority of laboratories perform testing for CD117 to a high standard.
The choice of primary antibody should be guided by the type of automation available to the laboratory.
Data availability statement
The data analyzed in this study is subject to the following licenses/restrictions: confidential quality assessment data. Requests to access these datasets should be directed to adodson@ukneqasiccish.org.
Ethics statement
Ethical approval was not required for the studies involving humans because Ethics not required for quality assurance study. The studies were conducted in accordance with the local legislation and institutional requirements. The human samples used in this study were acquired from Donnations from NHS hospital sample archives. Written informed consent to participate in this study was not required from the participants or the participants’ legal guardians/next of kin in accordance with the national legislation and the institutional requirements.
Author contributions
AD wrote the manuscript and analysed the data. SP gathered the assessment data and contributed to the writing. All authors contributed to the article and approved the submitted version.
Acknowledgements
The authors would like to acknowledge the contributions made to this work by the UK NEQAS ICC & ISH expert Assessors, and in particular Dr. Mark Pitt who has the pathologist lead for our CD117 in GIST Module for many years. The authors also acknowledge all the participating laboratories without whose data this study would not have been possible.
Conflict of interest
The authors(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Generative AI statement
The author(s) declared that generative AI was not used in the creation of this manuscript.
Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.
ReferencesSarlomo-RikalaMKovatichAJBaruseviciusAMiettinenM. CD117: a sensitive marker for gastrointestinal stromal tumors that is more specific than CD34. Mod Pathol (1998) 11(8):728–34.9720500MiettinenMLasotaJ. Gastrointestinal stromal tumors--definition, clinical, histological, immunohistochemical, and molecular genetic features and differential diagnosis. Virchows Arch (2001) 438(1):1–12. 10.1007/s00428000033811213830HeinrichMCRubinBPLongleyBJFletcherJA. Biology and genetic aspects of gastrointestinal stromal tumors: KIT activation and cytogenetic alterations. Hum Pathol (2002) 33(5):484–95. 10.1053/hupa.2002.12412412094373QasimHAbu ShugaerMAwawdehANDawaymehTKhattabKAl-OweiwiMGastrointestinal stromal tumors: histopathological spectrum, molecular subtypes, and implications for targeted therapy. Cureus (2026) 18(1):e101180. 10.7759/cureus.10118041664736MiettinenMLasotaJ. Histopathology of gastrointestinal stromal tumor. J Surg Oncol (2011) 104(8):865–73. 10.1002/jso.2194522069171YamaguchiMTateGEndoYMiyakiM. Immunohistochemistry and c-kit gene analysis in determining malignancy in gastrointestinal stromal tumors. Hepatogastroenterology (2003) 50(53):1431–5.14571755de SilvaCMReidR. Gastrointestinal stromal tumors (GIST): C-kit mutations, CD117 expression, differential diagnosis and targeted cancer therapy with Imatinib. Pathol Oncol Res (2003) 9(1):13–9. 10.1007/BF0303370812704441NovelliMRossiSRodriguez-JustoMTanierePSeddonBToffolattiLDOG1 and CD117 are the antibodies of choice in the diagnosis of gastrointestinal stromal tumours. Histopathology (2010) 57(2):259–70. 10.1111/j.1365-2559.2010.03624.x20716168Rios-MorenoMJJaramilloSPereira GallardoSVallejoAMoraMGarcía-EscuderoAGastrointestinal stromal tumors (GISTs): CD117, DOG-1 and PKCtheta expression. Is there any advantage in using several markers?Pathol Res Pract (2012) 208(2):74–81. 10.1016/j.prp.2011.11.00622197035AshmanLKGriffithR. Therapeutic targeting of c-KIT in cancer. Expert Opin Investig Drugs (2013) 22(1):103–15. 10.1517/13543784.2013.74001023127174DematteoRPHeinrichMCEl-RifaiWMDemetriG. Clinical management of gastrointestinal stromal tumors: before and after STI-571. Hum Pathol (2002) 33(5):466–77. 10.1053/hupa.2002.12412212094371RadfordIR. Imatinib. Novartis. Curr Opin Investig Drugs (2002) 3(3):492–9.12054102VerweijJvan OosteromABlayJYJudsonIRodenhuisSvan der GraafWImatinib mesylate (STI-571 glivec, gleevec) is an active agent for gastrointestinal stromal tumours, but does not yield responses in other soft-tissue sarcomas that are unselected for a molecular target. Results from an EORTC soft tissue and bone sarcoma group phase II study. Eur J Cancer (2003) 39(14):2006–11.12957454DorfmanDMBuiMMTubbsRRHsiEDFitzgibbonsPLLindenMDThe CD117 immunohistochemistry tissue microarray survey for quality assurance and interlaboratory comparison: a college of American pathologists cell markers committee study. Arch Pathol Lab Med (2006) 130(6):779–82. 10.5858/2006-130-779-TCITMS16740027RogeRBzorekMNielsenOVybergM. NordiQC assessments of CD117 immunoassays. Appl Immunohistochem Mol Morphol (2019) 27(2):87–91. 10.1097/PAI.000000000000071430499819UK_NEQAS_ICC_&_ISH. UK national external quality assessment scheme for immunocytochemistry and in-situ hybridisation (2026). Available online at: https://ukneqasiccish.org/ (Accessed: April 27, 2026).Roche_Diagnostics. Instructions for use: PATHWAY® Anti-c-KIT (9.7) primary antibody, catalog number 790-2951 (2008). Available online at: https://elabdoc-prod.roche.com/eLD/api/downloads/28c95b17-7133-ea11-fa90-005056a772fd?countryIsoCode=XG (Accessed: April 28, 2026).Diagnostic_Biosystems. Instructions for use: CD117/C-kit rabbit monoclonal antibody, RMAB119 (2024). Available online at: https://dbiosys.com/wp-content/uploads/2024/11/DS-0054-E.pdf (Accessed: April 28, 2026).Roche_Diagnostics. Instructions for use: anti-CD117 (EP10) rabbit monoclonal primary antibody (2022). Available online at: https://elabdoc-prod.roche.com/eLD/api/downloads/020888e8-0c15-ee11-2091-005056a71a5d?countryIsoCode=XG (Accessed: April 28, 2026).Leica_Biosystems. Instructions for use: rabbit monoclonal antibody c-kit oncoprotein (CD117), product code: NCL-L-CD117-032 (2018). Available online at: https://s3-qrd-prd-docs.s3.eu-west-1.amazonaws.com/leica/documents/8a3e223d-90d0-40ae-a15b-77a34ba79b03.pdf?response-content-disposition=filename%3D%22CD117-032-L-CE.pdf%22 (Accessed: April 28, 2026).Leica-Biosystems. Instructions for use: BOND ready-to-use primary antibody CD117 (EP10), catalog no: PA0007 (2022). Available online at: https://s3-qrd-prd-docs.s3.eu-west-1.amazonaws.com/leica/documents/f737ba71-161e-4ccd-aa9d-a2d0eb2f7f68.pdf?response-content-disposition=filename%3D%22PA0007.pdf (Accessed: April 28, 2026).Agilent_Dako. Instructions for use: polyclonal rabbit anti-human CD117, c-kit code A4502 (2020). Available online at: https://www.agilent.com/cs/library/packageinsert/public/304864EN_03.pdf (Accessed: April 28, 2026).HornickJLFletcherCD. Validating immunohistochemical staining for KIT (CD117). Am J Clin Pathol (2003) 119(3):325–7. 10.1309/ej3ry45qcypukqg412645332Cell_Marque. Instructions for use: CD117, c-kit (YR145) rabbit monoclonal primary antibody (2026). Available online at: https://www.cellmarque.com/cmsial-ifu/ASR11731070.pdf (Accessed: April 28, 2026).WongNAMeleghZ. Antigen retrieval and primary antibody type affect sensitivity but not specificity of CD117 immunohistochemistry. Histopathology (2009) 54(5):529–38. 10.1111/j.1365-2559.2009.03272.x19413635