The Clinical Characteristics of Patients and Sequencing Data
Three hundred and eleven patients with advanced cancer presented to Newcastle ECMC and had tumour profiling completed September 2017 – December 2020. Median age was 63 years (range 19 – 97), with a female predominance (n = 182, 58.5%). ‘FOL original’ was the most frequent sequencing technique (n = 151, 48.6%), followed by ‘FOL CDx’ and a Qiagen comprehensive cancer panel (PROSPECT-NE), equally (n = 80, 25.7%). Tumour types identified at ≥5% in the cohort were lung (n = 131, 42.1%), colorectal (n = 44, 14.1%), breast (n = 36, 11.6%) and prostate (n = 18, 5.6%) (Table 1). At least one genomic finding was identified in the majority of samples (n = 260, 84%).
Table 1 Profiled patients demographics Northern England September 2017 – December 2020
Variable
|
n = 311 (%)
|
Sex
M
F
|
129 (41.5)
182 (58.5)
|
Age (years)
Median (Range)
|
63 (19 – 97)
|
Sequencing Method
Foundation One® FOL Original
Foundation One® FOL CDx
Qiagen comprehensive cancer panel (PROSPECT-NE)
|
151 (48.6)
80 (25.7)
80 (25.7)
|
Tumour Profiling
Solid tumour tissue
ctDNA
|
80 (25.7)
231 (74.3)
|
Tumour Type
Lung
Colorectal
Breast
Prostate
Pancreatic
Cervical
Oesophagogastric
Ovarian
Cancer of Unknown Primary (CUP)
Bladder
Other[a]
|
131 (42.1)
44 (14.1)
36 (11.6)
18 (5.6)
12 (3.9)
9 (2.9)
8 (2.6)
5 (1.6)
5 (1.6)
5 (1.6)
38 (12.2)
|
[a] Other – Tumour types found in ≤4 patients: Adrenocortical, Appendiceal, Cholangiocarcinoma, Endometrial, Eccrine adenocarcinoma, Gastrointestinal stromal cell tumour (GIST), Liver, Renal, Sarcoma, Thymic, Vulval, No active malignancy.
Regional Prevalence of Mutations Compared to The Cancer Genome Atlas (TCGA) Genomic Data Commons (GDC) Data Portal
Significant differences in prevalence of mutations compared to TCGA were identified. Mutations identified at ≥5% in the cohort were compared (Table 2). Overall, 1267 lung cancer cases were retrieved from TCGA which revealed a higher relative prevalence of EGFR (n = 30, 22.9% vs n = 148, 11.7%, P = 0.001), RB1 (n = 17, 13.0% vs n = 82, 6.5%, P = 0.01) and CHEK2 (n = 11, 8.4 % vs n = 20, 1.6%, P = 0.000001) mutations in the Northern England population. Six hundred and eleven colorectal cancer cases were retrieved which did not reveal any significantly different prevalence in mutations. 1306 breast cancer cases retrieved also revealed a high relative prevalence of RB1 (n = 5, 13.9% vs n = 33, 2.5%, P = 0.0002) in the Northern England population. Five hundred and twenty-seven cases of prostate cancer retrieved from TCGA GDC revealed significantly different prevalence of mutations across all mutations identified ≥5% including TP53 (n = 13, 72.2% vs n = 70, 13.3%, P = 0.00001), AR (n = 6, 33.3% vs n = 4, 0.8%, P = 0.00001) and PTEN (n = 5, 27.8% vs n = 19, 3.6%, P = 0.00002).
Table 2 Mutations in common tumour types identified in Northern England compared to Cancer Genome Atlas
Mutations commonly identified in Newcastle ECMC
|
Northern England
|
Cancer Genome Atlas
|
P
(Chi-Squared)
|
Lung
|
|
n = 131 (%)
|
n = 1267 (%)
|
|
TP53
|
72 (55.0)
|
887 (69.0)
|
0.11
|
EGFR
|
30 (22.9)
|
148 (11.7)
|
0.001
|
KRAS
|
23 (17.6)
|
208 (16.4)
|
0.78
|
RB1
|
17 (13.0)
|
82 (6.5)
|
0.01
|
STK11
|
15 (11.5)
|
117 (9.2)
|
0.46
|
ATM
|
14 (10.7)
|
104 (8.2)
|
0.38
|
CHEK2
|
11 (8.4)
|
20 (1.6)
|
0.000001
|
PIK3CA
|
11 (8.4)
|
104 (8.2)
|
0.95
|
PTEN
|
10 (7.6)
|
89 (7.0)
|
0.81
|
DNMT3A
|
9 (6.9)
|
44 (3.5)
|
0.07
|
Colorectal
|
|
n = 44 (%)
|
n = 610 (%)
|
|
TP53
|
33 (75.0)
|
386 (63.3)
|
0.48
|
APC
|
28 (63.6)
|
486 (79.7)
|
0.37
|
KRAS
|
22 (50.0)
|
255 (41.8)
|
0.51
|
PIK3CA
|
8 (18.2)
|
165 (27.0)
|
0.31
|
MSH6
|
3 (6.8)
|
42 (6.9)
|
0.77
|
BRAF
|
3 (6.8)
|
86 (14.1)
|
0.22
|
NRAS
|
3 (6.8)
|
32 (5.2)
|
0.67
|
ERBB2
|
2 (4.5)
|
39 (6.4)
|
0.64
|
Breast
|
|
n = 36 (%)
|
n = 1306 (%)
|
|
TP53
|
13 (36.1)
|
473 (36.2)
|
0.99
|
PIK3CA
|
10 (27.8)
|
435 (33.3)
|
0.62
|
RB1
|
5 (13.9)
|
33 (2.5)
|
0.0002
|
PTEN
|
4 (11.1)
|
89 (6.8)
|
0.35
|
ATM
|
3 (8.3)
|
38 (2.9)
|
0.08
|
Prostate
|
|
n = 18 (%)
|
n = 527 (%)
|
|
TP53
|
13 (72.2)
|
70 (13.3)
|
0.00001
|
AR
|
6 (33.3)
|
4 (0.8)
|
0.00001
|
PTEN
|
5 (27.8)
|
19 (3.6)
|
0.00002
|
PIK3CA
|
3 (16.7)
|
13 (2.5)
|
0.00001
|
TMPRSS2
|
3 (16.7)
|
9 (1.7)
|
0.000095
|
AKT1
|
2 (11.1)
|
3 (0.6)
|
0.000013
|
CTNNB1
|
2 (11.1)
|
12 (2.3)
|
0.029
|
NF1
|
2 (11.1)
|
2 (0.4)
|
0.000001
|
The Utility of Experimental Cancer Medicine Centre (ECMC) Cancer Research UK (CRUK) Trial Finder
EC Trial Finder demonstrated significantly different utility than commercial sequencing reports in identifying trials for mutations identified at ≥5% prevalence (P = 0.007) in lung, colorectal, breast and prostate cancer (Figure 1 and Appendix 1). Sequencing reports identified actionable mutations and suggested targeted trials in the majority (n = 23, 74.1%) of 31 mutations explored across these tumour types compared to less than half (n = 14, 45.2%) utilising EC Trial Finder.
Examples of where commercial reports might significantly differ from either the national trial finder or the opinion of the molecular tumour board are highlighted by the CHEK2 abnormalities in the 8 patients with lung cancer. These included:
1. Describing studies available in different countries (for example NCT04123366 enrolling in Europe but not specifically the UK, NCT02498613 enrolling in Canada and the USA). Whilst patient choice is key, there will be additional barriers in patients enrolling on studies in different countries where they may not be eligible for standard treatment costs. In addition, this may cause harm in terms of financial toxicity and time away from work, family and friends. This was a frequent cause of discrepancy across tumour types.
2. The suggestion of basket studies looking for efficacy through synthetic lethality with either PARP or ATR inhibitors (for example NCT03742895). There is no emerging evidence to support synthetic lethality in this setting, and the identified basket studies were not enrolling patients with those specific genetic abnormalities. This may not be obvious to a commercial companies and trial sponsors as inclusion criteria on clinical trials.gov only states that patients must have mutations in the homologous recombination repair pathway.
3. Highlighting studies with genomically unselected use of a PARP inhibitors: For example, maintenance treatment following first-line treatment with chemotherapy and immunotherapy combinations (NCT03976362). Here not only are genomic aberrations not a key inclusion criteria, but most patients will not be eligible on the basis of where they are in the disease course as this is only suitable for previously untreated stage IV patients.
4. Description studies with the incorrect tumour type: For example, highlighting studies recruiting patients with small-cell lung cancer rather than non-small-cell lung cancer (NCT02769962).
These reasons for discrepancies were not mutually exclusive and for some studies more than one reason of those above lead to the study not being considered by either the national trial finder or the local MTB. Similar issues were encountered with other genomics abnormalities and disease types. For example, highlighting studies of PI3 kinase inhibitors in any tumour with an abnormality in a component of this pathway, regardless of the exact gene under question, or studies in different tumour types (Appendix 1).
Summary figure of trial exploration for prevalent (≥5%) mutations in four most common tumour types. PROSPECT-NE MTB results were retrospectively interrogated using electronic records and classified as potentially actionable by MTB. Potentially targeted trials were recorded. Foundation One® sequencing reports were retrospectively reviewed. Mutations were recorded as actionable, and trials recorded as YES if a report suggested a matched trial. All mutations were processed by tumour type using EC Trial Finder (November 2021) and recorded as actionable if trial modalities were suggested in that tumour type and mutation. Trials were recorded as YES if open and whether they were ‘all comer’ or specific to tumour type. EC Trial Finder demonstrated significantly different utility than sequencing reports in identifying trials for common mutations identified (P = 0.007) (McNemar’s). Significance testing criteria was not fulfilled for MTB results.