Biology and targetsPancreatic ductal adenocarcinoma

Cancer-associated fibroblasts

88% of 3,594 targeted-panel samples (Singhi 2019); 2,217 of 2,483, 89.3%, in the Caris cohort (Philip 2022); more than 90% of 109 microdissected exomes (Witkiewicz 2015); cBioPortal: 2,188 of 2,336, 93.7%, in pdac_msk_2024; 361 of 395, 91.4%, in pancreas_msk_2024; 135 of 140, 96.4%, in paad_cptac_2021; 344 of 383, 89.8%, in paad_qcmg_uq_2016; 100 of 109, 91.7%, in paad_utsw_2015; 94 of 99, 94.9%, in paad_icgc. A subset of tumours carries multiple KRAS mutations, some biallelic (Cancer Genome Atlas 2017); mutant KRAS copy gains were recorded in 296 of 853 dosage-annotated pdac_msk_2024 samples (2 or more extra copies in 201, 3 or more in 95).

G12D ~40%, G12V ~30%, G12R ~15%, G12C ~1-2%

cBioPortal: 1,778 of 2,336, 76.1%, in pdac_msk_2024; 286 of 395, 72.4%, in pancreas_msk_2024; 105 of 140, 75.0%, in paad_cptac_2021; 253 of 383, 66.1%, in paad_qcmg_uq_2016; 55 of 109, 50.5%, in paad_utsw_2015; 107 of 179, 59.8%, in paad_tcga_pan_can_atlas_2018; 33 of 99 in paad_icgc on 2012 exome depth. Hotspots R175H, R248Q, R248W, R273H and R273C recur. 44.5% of KRAS wild-type tumours (Philip 2022); the squamous subtype is enriched for TP53 and KDM6A mutation (Bailey 2016).

Approximate; heterogeneous

cBioPortal: mutation in 546 of 2,336, 23.4%, and deep deletion in 359, 15.4%, in pdac_msk_2024; mutation 35 of 179, 19.6%, and deep deletion 52 of 183, 28.4%, in paad_tcga_pan_can_atlas_2018; mutation 6 and deep deletion 40 of 109, 42.2%, in paad_utsw_2015; mutation 93 of 395, 23.5%, and deep deletion 19 in pancreas_msk_2024; mutation 71 of 383, 18.5%, in paad_qcmg_uq_2016 (no copy-number profile); 29 of 140, 20.7%, in paad_cptac_2021. CDKN2B was co-deleted in 333 of 2,336 (pdac_msk_2024) and 50 of 183 (TCGA); MTAP deep deletion in 40 of 183, 21.9% (TCGA) and 34 of 109, 31.2% (UTSW), 51 of 2,336 on the MSK panel.

Wide range across series

889 of 2,188 KRAS mutation records, 40.6%, in pdac_msk_2024 (the study's own KRAS_VARIANT attribute: 892 G12D, 706 G12V, 352 G12R, 111 Q61H, 37 Q61R, 26 G12C, 70 other); 136 of 344, 39.5%, in paad_qcmg_uq_2016; 61 of 135, 45.2%, in paad_cptac_2021; 39 of 95, 41.1%, in paad_icgc; 49 of 119, 41.2%, in paad_tcga_pan_can_atlas_2018; 37 of 100, 37.0%, in paad_utsw_2015 (cBioPortal). Among 219 advanced patients genotyped on EUS-guided biopsies, G12D 73, G12V 53, G12R 21 (Bournet 2016).

709 of 2,188, 32.4%, in pdac_msk_2024; 108 of 344, 31.4%, in paad_qcmg_uq_2016; 40 of 135, 29.6%, in paad_cptac_2021; 35 of 95, 36.8%, in paad_icgc; 33 of 119, 27.7%, in paad_tcga_pan_can_atlas_2018; 40 of 100, 40.0%, in paad_utsw_2015 (cBioPortal, mutation records). 53 of 147 codon-12 mutations in 219 advanced patients (Bournet 2016).

cBioPortal: mutation in 512 of 2,336, 21.9%, and deep deletion in 102, 4.4%, in pdac_msk_2024 (R361H 30, R361C 26, R445* 15); 86 of 383, 22.5%, in paad_qcmg_uq_2016; mutation 37 of 179, 20.7%, and deep deletion 23 of 183, 12.6%, in paad_tcga_pan_can_atlas_2018; mutation 21 and deep deletion 32 of 109 in paad_utsw_2015; 68 of 395, 17.2%, plus 10 deletions in pancreas_msk_2024; 24 of 140, 17.1%, in paad_cptac_2021. TGFBR2 mutation in 91 of 2,336, 3.9%, and 18 of 383, 4.7%; TGFBR1 64 of 2,336; ACVR1B 11 of 383 (cBioPortal). SMAD4 and TGFBR2 mutations were largely restricted to the invasive carcinoma in cyst progression (Noe 2020).

354 of 2,188, 16.2%, in pdac_msk_2024; 61 of 344, 17.7%, in paad_qcmg_uq_2016; 23 of 135, 17.0%, in paad_cptac_2021; 16 of 95, 16.8%, in paad_icgc; 25 of 119, 21.0%, in paad_tcga_pan_can_atlas_2018; 12 of 100, 12.0%, in paad_utsw_2015 (cBioPortal, mutation records). G12R is almost unique to pancreatic cancer among KRAS-driven tumours and carried the longest survival of the codon-12 alleles, 14 months, in 219 advanced patients (Bournet 2016).

266 of 2,483, 10.7% (Philip 2022); 12% of 3,594 (Singhi 2019); 148 of 2,336, 6.3%, in pdac_msk_2024; 39 of 383, 10.2%, in paad_qcmg_uq_2016; 5 of 99 by the study's KRAS_MUTATION attribute in paad_icgc; 34 of 395, 8.6%, in pancreas_msk_2024 (cBioPortal). Among KRAS wild-type tumours: TP53 mutated in 44.5%, BRAF in 13.0%, fusions of BRAF 6.6%, FGFR2 5.2%, ALK 2.6%, RET 1.3% and NRG1 1.3%, amplification of FGF3 3%, ERBB2 2.2%, FGFR3 1.8%, NTRK 1.8% and MET 1.3%; MSI-high 4.7% against 0.7% and TMB-high 4.5% against 1% in KRAS-mutant disease (Philip 2022). KRAS wild-type TCGA tumours carried GNAS, BRAF, CTNNB1 and other RAS-pathway alterations (Cancer Genome Atlas 2017); all 4 of 17 young adults (Heining 2018) and all 3 of 47 patients (Jones 2019) with wild-type tumours carried a kinase or NRG1 fusion on whole-genome sequencing.

cBioPortal high-level amplification: 23 of 183, 12.6%, in paad_tcga_pan_can_atlas_2018; 13 of 109, 11.9%, in paad_utsw_2015; 98 of 2,336, 4.2%, in pdac_msk_2024; 11 of 395, 2.8%, in pancreas_msk_2024. MYC amplification was uniquely associated with poor outcome and the adenosquamous subtype among 109 microdissected cancers (Witkiewicz 2015); squamous-feature cancers showed intercellular heterogeneity for MYC amplification (Hayashi 2020); KDM6A loss activates a MYC super-enhancer (Andricovich 2018).

cBioPortal: 204 of 2,336, 8.7%, in pdac_msk_2024; 35 of 395, 8.9%, in pancreas_msk_2024; 29 of 383, 7.6%, in paad_qcmg_uq_2016; 9 of 179, 5.0%, in paad_tcga_pan_can_atlas_2018; 8 of 140 in paad_cptac_2021; 7 of 109 in paad_utsw_2015. KMT2C 73, KMT2D 100, SMARCA4 52, PBRM1 32 and ARID2 30 of 2,336 in the same study. Chromatin-regulating gene mutations (MLL, MLL2, MLL3, ARID1A) in 20% of 101 patients were associated with improved survival (Sausen 2015); ARID1A was a significantly mutated gene in 99 exomes (Biankin 2012) and a recurrent gene in TCGA (Cancer Genome Atlas 2017).

167 of 2,188, 7.6% (Q61H 107, Q61R 38, Q61L 14, Q61K 8), in pdac_msk_2024; 23 of 344, 6.7% (Q61H 19), in paad_qcmg_uq_2016; 8 of 135, 5.9%, in paad_cptac_2021; 5 of 95, 5.3%, in paad_icgc; 8 of 119, 6.7%, in paad_tcga_pan_can_atlas_2018; 7 of 100, 7.0%, in paad_utsw_2015 (cBioPortal, mutation records). Codon 61 alleles were selectively associated with improved survival among 109 microdissected cancers (Witkiewicz 2015).

cBioPortal: 142 of 2,336, 6.1%, in pdac_msk_2024; 30 of 395, 7.6%, in pancreas_msk_2024; 11 of 179, 6.1%, in paad_tcga_pan_can_atlas_2018; 21 of 383, 5.5%, in paad_qcmg_uq_2016; 9 of 140, 6.4%, in paad_cptac_2021; 6 of 109 in paad_utsw_2015. Mutated in 6 of 8 IPMNs and 3 of 8 MCNs, establishing it as a suppressor of both (Wu 2011, PNAS); RNF43 alterations were largely confined to the non-invasive lesions in cyst progression (Noe 2020); a recurrent gene in TCGA (Cancer Genome Atlas 2017) and part of the WNT pathway of the ten (Bailey 2016).

cBioPortal: 91 of 2,336, 3.9%, in pdac_msk_2024; 12 of 383, 3.1%, in paad_qcmg_uq_2016; 7 of 179, 3.9%, in paad_tcga_pan_can_atlas_2018; 12 of 395, 3.0%, in pancreas_msk_2024. Named a new candidate driver by structural variation in 100 whole genomes (Waddell 2015); the squamous subtype is enriched for KDM6A mutation (Bailey 2016); KDM6A loss induced squamous-like, metastatic tumours through super-enhancer activation of TP63, MYC and RUNX3 and conferred BET inhibitor sensitivity in mice (Andricovich 2018).

cBioPortal: 76 of 2,336, 3.3% (R201H 39, R201C 21, R201S 3), in pdac_msk_2024; 7 of 179, 3.9% (R201C 4, R201H 2), in paad_tcga_pan_can_atlas_2018; 10 of 383, 2.6%, in paad_qcmg_uq_2016; 9 of 395, 2.3%, in pancreas_msk_2024; 5 of 109 in paad_utsw_2015; GNAS was a recurrent gene in the TCGA analysis and KRAS wild-type tumours carried GNAS among their drivers (Cancer Genome Atlas 2017). GNAS mutations in 66% of 132 IPMNs, KRAS or GNAS in 96%, carried into the invasive cancer in 7 of 8 IPMN-associated carcinomas and absent from other cysts and from non-IPMN cancers (Wu 2011); 27 of 456 QCMG tumours were IPMN with invasion (cBioPortal histology).

cBioPortal high-level amplification: 9 of 183, 4.9%, in paad_tcga_pan_can_atlas_2018; 12 of 109, 11.0%, in paad_utsw_2015; 28 of 2,336, 1.2%, plus 18 mutations in pdac_msk_2024; 4 of 395 in pancreas_msk_2024. Focal amplifications of ERBB2, MET, FGFR1, CDK6, PIK3R3 and PIK3CA were found at low individual prevalence in 100 whole genomes (Waddell 2015); ERBB2 amplification in 2.2% of 266 KRAS wild-type tumours (Philip 2022).

cBioPortal: mutation in 22 of 2,336, 0.9% (N486_P490del 6, V600E 3, V600R 2, T599dup 2), plus 7 fusions (SND1-BRAF 3) in pdac_msk_2024; 3 of 109, all V600E, in paad_utsw_2015, where BRAF mutations were mutually exclusive with KRAS and conferred vemurafenib sensitivity in models (Witkiewicz 2015); 2 of 179 in paad_tcga_pan_can_atlas_2018 and 4 of 395 (V600E 2) in pancreas_msk_2024. 13.0% of 266 KRAS wild-type tumours carried a BRAF mutation and 6.6% a BRAF fusion (Philip 2022). Two of 71 advanced patients carried oncogenic in-frame BRAF deletions, the first clinical evidence that they respond to MAPK pathway inhibition (Aguirre 2018).

2.3% of 3,030 against 0.37% of controls, odds ratio 5.71 (Hu 2018); 10 of 854, 1.2% (Shindo 2017); ATM among the new significantly mutated genes in 99 exomes (Biankin 2012). Tumour sequencing: 69 of 2,336, 3.0%, in pdac_msk_2024; 14 of 383, 3.7%, in paad_qcmg_uq_2016; 8 of 179, 4.5%, in paad_tcga_pan_can_atlas_2018 (cBioPortal). Germline ATM carriers with one affected first-degree relative became eligible for CAPS surveillance in 2020 (Goggins 2020).

1.9% of 3,030 patients against 0.3% of controls, odds ratio 6.20 (Hu 2018); 12 of 854 apparently sporadic patients, 1.4% (Shindo 2017); 4 of 289 resected patients (Yurgelun 2019); 49 of 71 BRCA carriers in a three-centre series were BRCA2 (Golan 2014). Tumour sequencing (germline and somatic together): 67 of 2,336, 2.9%, in pdac_msk_2024; 8 of 383, 2.1%, in paad_qcmg_uq_2016; 11 of 395, 2.8%, in pancreas_msk_2024; 2 of 179 in paad_tcga_pan_can_atlas_2018 (cBioPortal). POLO screened 3,315 metastatic patients to randomise 154 germline BRCA1/2 carriers (Golan 2019).

About 1 to 2% of pancreatic cancers (Strickler 2023); 28 of 2,188 KRAS mutation records, 1.3%, and 26 of 2,194 samples with a KRAS_VARIANT call, 1.2%, in pdac_msk_2024; 6 of 344, 1.7%, in paad_qcmg_uq_2016; 4 of 100, 4.0%, in paad_utsw_2015; 3 of 361 in pancreas_msk_2024; 1 of 135 in paad_cptac_2021; 1 of 119 in paad_tcga_pan_can_atlas_2018 (cBioPortal). CodeBreaK 100 treated 38 previously treated patients: 8 centrally confirmed responses, 21%, progression-free survival 4.0 months, overall survival 6.9 months (Strickler 2023); KRYSTAL-1 gave 7 of 21 pancreatic responses, 33.3%, to adagrasib (Bekaii-Saab 2023).

7 of 833, 0.8%, all with Lynch syndrome, 4 of 7 benefiting from checkpoint blockade (Hu 2018, Clin Cancer Res); MSI-high and/or TMB-high in 0.5% of 2,563 assessed (Singhi 2019); 1 to 2% across 34 studies and 8,323 patients, associated with medullary and colloid histology and KRAS/TP53 wild-type background (Luchini 2021); 4.7% of KRAS wild-type against 0.7% of KRAS-mutant tumours (Philip 2022); cBioPortal: MSI_TYPE unstable in 6 of 1,119 reported samples and MSIsensor score 10 or more in 10 of 2,314 in pdac_msk_2024; 2 of 323 reported in pancreas_msk_2024; no TCGA sample reached MANTIS 0.4. Germline MLH1 0.13% of 3,030, odds ratio 6.66 (Hu 2018, JAMA).

0.6% of 3,030 against 0.2% of controls, odds ratio 2.58 (Hu 2018); 3 of 854, 0.4% (Shindo 2017); 3 of 289 (Yurgelun 2019); 21 of 71 BRCA carriers (Golan 2014). Tumour sequencing: 16 of 2,336, 0.7%, in pdac_msk_2024; 5 of 383 in paad_qcmg_uq_2016; 2 of 179 in paad_tcga_pan_can_atlas_2018 (cBioPortal).

cBioPortal structural variants: 6 of 2,336 samples, 0.26% (ATP1B1-NRG1 2, CD44-NRG1 1, NOTCH2-NRG1 1), in pdac_msk_2024 and 2 of 184, 1.1% (ATP1B1-NRG1, UBXN8-NRG1), in paad_tcga_pan_can_atlas_2018; 1.3% of 266 KRAS wild-type tumours (Philip 2022); recurrent in KRAS wild-type young adults, with ERBB inhibition giving remissions of liver metastases in 2 patients (Heining 2018); present in all 3 KRAS wild-type tumours among 47 whole genomes, with rapid responses to afatinib in 2 (Jones 2019). Zenocutuzumab gave responses in 15 of 36 pancreatic patients, 42%, in eNRGy (Schram 2025).

cBioPortal structural variants: 9 of 2,336, 0.4% (ETV6-NTRK3 3, CTRC-NTRK1 2, EML4-NTRK1, COP1-NTRK1), in pdac_msk_2024; 2 of 184 (CEL-NTRK1, EML4-NTRK3) in paad_tcga_pan_can_atlas_2018; NTRK amplification 1.8% of 266 KRAS wild-type tumours (Philip 2022). A CTRC-NTRK1 fusion cancer responded to larotrectinib for 6 months before resistance (O'Reilly and Hechtman 2019).

2 of 854 germline, 0.2% (Shindo 2017); listed among the double-strand repair genes in 21 of 289 resected patients (Yurgelun 2019) and the 122 germline carriers of 615 (Lowery 2018); 15 of 2,336 tumours, 0.6%, in pdac_msk_2024 and 4 of 395 in pancreas_msk_2024 (cBioPortal). Six germline PALB2 carriers among 42 evaluable patients responded to maintenance rucaparib at 50%, 3 of 6 (Reiss 2021). Genomic instability co-segregated with BRCA1, BRCA2 or PALB2 inactivation in 100 whole genomes (Waddell 2015).

5 of 3,170 profiled cancers, 0.16%, all KRAS wild-type and all under 50, where ALK fusions were 1.3% of tumours in patients under 50; 3 of 4 treated with an ALK inhibitor had stable disease, radiographic response or CA 19-9 normalisation (Singhi 2017); 1 of 2,336 (ASAP2-ALK) in pdac_msk_2024 (cBioPortal); 2.6% of 266 KRAS wild-type tumours (Philip 2022).

cBioPortal structural variants: 4 of 2,336, 0.17% (FGFR2-CAT 2, FGFR2-SHTN1), in pdac_msk_2024; 5.2% of 266 KRAS wild-type tumours (Philip 2022). Other receptor kinase fusions in pdac_msk_2024: ROS1 5 (GOPC-ROS1), RAF1 4, RET 2 (NCOA4-RET), MET 2 (KANK1-MET), ERBB2 2, EGFR 3, FGFR1 2, FGFR3 1 (cBioPortal).