RAIDS atlas of significant genetic and protein biomarkers in cervical cancer

Abstract

Loss of function in epigenetic acting genes together with driver alterations in the PIK3CA pathway have been shown significantly associated with poor outcome in cervical squamous cell cancer. More recently, a CoxBoost analysis identified 16 gene alterations and 30 high level activated proteins to be of high interest, due to their association with either good or bad outcome, in the context of treatment received by chemoradiation. The objectives here were to review and confirm the significance of these molecular alterations as suggested by literature reports and to pinpoint alternate treatments options for poor-responders to chemoradiation.

INTRODUCTION

Combining DNA repair with microtubule interference appeared more effective than platinum alone for treatment naïve cervical cancers (CCs), while also resulting in higher levels of tumor-infiltrating T-cells [1]. Significant efforts are made worldwide to include patients who failed “standard therapies” into biomarker-based platform trials since predictions of response in molecularly undefined tumors remain uncertain. While addition of immune checkpoint blockade (ICB) in second line treatment of metastatic CC has shown promise [2], the combination of ICB with chemoradiotherapy in treatment naïve patients did not improve progression-free survival (PFS) in the CALLA trial (NCT03830866). BioRAIDs (NCT02428842), a prospectively collected CC dataset (n=419), was part of the EU funded consortium “RAIDS” (Rational molecular Assessments and Innovative Drugs selection) whose aims were to develop innovative tools, models, and new standards for CC treatment. Analysis of genetic (whole exome sequencing) and protein (reverse phase protein array, RPPA) data from this prospectively sampled CC patient population, detected number of genetic and protein driver dependencies which correlated with PFS [3,4]. To corroborate our results, the literature was reviewed for independent evidence on the prognostic value of the genetic or protein markers which had been associated with response to chemoradiation by us. Additionally, alternate treatment options to address high-ranking protein alterations or specific genetic alterations that have been associated with poor outcome following chemoradiation, are discussed. Innovative single or combined treatments, targeted to specific pathways (epigenetic, metabolic, immune, tyrosine kinase activation) or human papillomavirus (HPV) viral antigens as well as treatments affecting protein degradation (proteasome inhibitors) have already shown promise and may be further assessed in platform trials following detailed tumor assessments by a molecular tumor board.

METHODOLOGY OF LITERATURE REVIEW

Genomic and proteomic assessments in the prospective study BioRAIDs had been carried out in one single batch, at one Institute. Antibodies had been validated by Western blot, thus avoiding batch effects and technical issues related to platform, mapping pipeline, sequencing depths, mutation calling, choice and quality control of antibodies.

Gain of function (GoF) or loss of function (LoF) in 16 genes together with overexpression, loss of expression or specific post translational modifications of 30 proteins of clinical relevance [3,4] were reviewed in the light of clinical and preclinical reports, perusing PubMed, cBioPortal, GeneCards. Focus was on squamous cell cancer but other non-gynecological and non-squamous-cell tumor types were also considered. Reports were limited to studies with >50 patients/human derived samples from the years 2010–2022.

The BioRAIDs CC analytical gene panel had focused on 100 clinically well-known oncogenes 203 tumor-suppressor genes and 306 genes classified as “of uncertain clinical relevance.” The genetic biomarkers refer to variants present in at least 5% of the CC tumor samples in the 146 first samples of the BioRAIDs population. Alterations were: single nucleotide variant, rearrangements, frameshifts, deletions, insertions as well as gene amplifications or bi-allelic loss of suppressor functions. Pathogenic gene variants [3] had been defined as follows: (i) for oncogenes: only mutations driving to GoF were considered (i.e., hotspots missense mutations, in-frame insertions/deletions/splicing described as oncogenic in the literature), (ii) for tumor suppressor genes (TSGs): only mutations driving to LoF were considered (i.e., truncating alterations [nonsense mutations, frameshift insertions/deletions/splicing]) or missense mutations; copy number alterations with focal amplifications for oncogenes and homozygous deletions for TSG or loss of heterozygosity when associated with a pathogenic variant. Mutations of known significance provided by the American Type Culture Collection database had been used as validations points. The detection of activated/inactivated or differentially phosphorylated proteins, from a RPPA analysis, had been based on 162 antibodies, which had all been previously validated on Western Blots in one experiment, including 156 patients and 20 colon cancer cell line. Eighty-nine patients with complete data sets in genomics and proteomics were included in the CoxBoost analysis.

Clinical trial numbers as listed in the ClinicalTrials.gov-beta database of potential interest in specific gene alterations or protein overexpression/modifications are listed in the tables.

LITERATURE ACCOUNT OF SALIENT GENES AND PROTEINS POTENTIALLY INVOLVED IN CC OUTCOME

1. Genetic alterations linked to poor outcome following chemoradiation

Cub and Sushi multiple domain 3 (CSMD3) (LoF)

CSMD3 encodes a large cell membrane protein involved in dendritic development and in complement regulation (Table 1) [6,7,8,9,10,11,12,13,14]. First identified in non-small cell lung cancer (NSCLC), LoF by somatic mutation was second in frequency only to p53 variants and has been suggested as a potential therapeutic target in pulmonary carcinosarcoma. CoxBoost analysis revealed this to be the most significant gene alteration associated with resistance to chemoradiation in CC [4], while receiver operating characteristic (ROC) analysis in small cell lung cancer (SCLC) revealed this to be the most significant gene alteration linked to etoposide resistance, a topoisomerase II inhibitor [5]. CSMD3 gene amplifications have also been reported and may lead to gain-of-function in other cancer types. CSMD3 alterations are found in 9% to 11% in CC.

Table 1. Alternative targets for genetic alterations resistant to DNA repair inhibition.

Gene name, alteration type - frequency		Physiopathology and treatment resistance	Alternate treatments, type of evidence, No. of clinical trials
CSMD3: LoF by MM, SM, TM, Amp		CSMD3 deletion enhanced airway epithelial cell growth and CSMD3 LoF mutations have been linked to etoposide resistance in small cell lung cancer.	Phase 3 & combination CT
	Cervix: TCGA CC - BioRAIDs 11%	ICB appeared beneficial in first line therapy in high-grade serous ovarian carcinomas with CSMD3 LoF [6].	Pembrolizumab: 1,845 CT
	Ovary: TCGA 26%		16 CT in “neoplasm cervix”
	Endometrium: TCGA 27%
UBR5 (alias EDD1): LoF by MM, SM, TM, Amp, TR		UBR5 LoF can affect several oncogenic pathways including p53, NF-κB, and DNA repair.	Phase 3 & combination trials
	Cervix: TCGA 6% - BioRAIDs 5%	Velcade (Bortezomib), a proteasome inhibitor demonstrated significant efficacy in the treatment of refractory MM and MCL with UBR5 LoF.	Bortezomib: 935 CT
	Ovary: TCGA 12%	E3 ligase modulators have been suggested in preclinical and early CT. Immune stimulatory effects have been demonstrated in preclinical assays by Iberdomide.	Ixazomib: 59 CT
	Endometrium: TCGA 16%		Lenalidomide: 1,053 CT
			Pomalidomide: 221 CT
ARID1A: LoF by MM, SM, TM, Amp, TR, DD		ARID1A modulates p53-mediated transcriptional regulation.	FDA approved biomarker
	Cervix: TCGA 7% - BioRAIDs 9%	Cancer cells lacking ARID1A (SWI/SNF) function are vulnerable to DNA damage and resistant to many treatments but sensitive to ICB [7]. ARID1A LoF is an FDA approved biomarker for ICB.	ICB for ARID1A gene mut.: 7 CT
	Ovary: TCGA 1,8%	EZH2 inhibitors are synthetically lethal to ARID1A LoF. Tazemetostat has been suggested, but may not fully suppress oncogenic activity [8].	Clinical/combination trials
	Endometrium: TCGA 45%		Tazemetostat: 37 CT
CREBBP (CBP): LoF by MM, SM, TM, Amp, IF-M		CREBBP knockdown enhances RAS/RAF/MEK/ERK signaling. Loss of CREBBP proteins has been shown to lead to cancer while increasing sensitivity to HDAC inhibition [9]. ICG-001 antagonized the CBP/β-catenin interaction, inhibiting tumorigenicity of pediatric glioma in a Wnt-independent manner [10]. Similarly, the small molecule E7386 attenuated Wnt signaling by blocking the interaction between β-catenin and CREB-binding protein (CBP).	Early & combination CT
	Cervix: TCGA 9% - BioRAIDs 8%		HDAC inhibitors: 667 CT
	Ovary: TCGA 8%		2 CT “neoplasm cervix uteri”
	Endometrium: TCGA 14%		ICG-001: 5 CT
			Nerve sparing/imaging in CC: 1 CT
			E7386: 4 CT
PIK3CA: GoF by activating mutations, MM, Amp, SM, TM, IF-M		PIK3CA activating mutations have been associated with resistance to chemoradiation, cisplatin, carbo-taxol and cetuximab. Many PIK3CA inhibitors and Pan-PI3K inhibitors: mTor inhibitors and downstream AKT inhibitors are in different stages of clinical investigation. One CT based on PIK3CA activating mutations using BKM120 was stopped. Concomitant action of PIK3 pathway and inhibition of the metabolic pathway via the glutaminase inhibitor, telaglenastat (CB-839) has appeared beneficial in preclinical CL studies. Clinical trials of telaglenastat have shown avorable systemic tolerance.	Clinical trials
	Cervix: TCGA39% - BioRAIDs 40%		Alpelisib: 92 CT
	Ovary: TCGA 22%		Not reimbursed in France
	Endometrium: TCGA 45%		Buparlisib: 90 CT
			Idelalisib: 70 CT
			Copanlisib: 65 CT
			Telaglenastat: 14 CT
			Taselisib: 9 CT
			Fimepinostat: 2 CT
TP53: LoF by MM, TM, DD		Mechanistic restoration for p53 LoF in tumors needs to be confirmed.	Early clinical trials
	Cervix: TCGA 9% - BioRAIDs 7%	A: Small molecules that restore wild-type conformation.	AMG 232: 8 CT
	Ovary: TCGA 92%	B: Action via inhibition of MDM2 [12]: Nutlins are cis-imidazoline analogues that inhibit the interaction between MDM2 and tumor suppressor p53 [11].	PRIMA-1: 2 CT
	Endometrium: TCGA 38%	C: GST fusion proteins.	RG7112: 3 CT
			GST fusion prot.: 2 CT
PTEN: LoF by MM, TM, DD, SM, A		PTEN LoF, estimated to be present in 30% of cancers is associated with resistance to treatment. Downstream targeting via AKT inhibitors has been suggested. Epigenetic reprogramming by bromodomain inhibitors (Molibresib) in combination with PARP inhibition (Talazoparib) suggested combinatorial drug synergy. Downregulation of homologous recombination-related proteins (PTEN, Chek2, Nbn, FancC) increased sensitivity to PARP inhibition [13].	Combination trials AKT inh.
	Cervix: TCGA 13% - BioRAIDs 9%		Capivasertib + Abiraterone in prostate cancer NCT04493853
	Ovary: TCGA 6%		Combination trial - other
	Endometrium: TCGA 68%		ZEN-3694 + Talazoparib NCT05327010
KMT2D (alias MLL2): LoF by MM, TM, DD, SM, A		KMT2D LoF mutation gives rise to significant genomic instability in vivo, partially explained by the fact that the ASCOM complex functions as a coactivator of the tumor suppressor p53. More than 12 Bromodomain inhibitors are in clinical trials, many in combination with other drugs such as ICB, +/− Azacitidine, Etoposide, Cisplatin, Talazoparib, Entinostat, Selumetinib depending on tumor type.	Single + Combination trials
	Cervix: TCGA15% - BioRAIDs 16%		Zen-3694: 12 CT
	Ovary: TCGA 2,8%		GSK 525762: 9 CT
	Endometrium: TCGA 29%		GSK 2820151: 9 CT
			CC 90010: 5 CT
			OTX015 Birabresib: 5 CT
			AZD 5153: 5 CT
			INCB 059872: 4 CT
NFE2L2 (NRF-2): GoF by MM, TM, DD, SM, A		NRF-2 is a master transcription factor in cellular redox hemostasis, essential for stem cell maintenance. Inhibitors of NRF-2 or its regulator KEAP1 in clinical development are reviewed in Dempke and Reck [14]. MSU38225 downregulates transcriptional activity of NRF-2. Brusatol causes Ubiquitin degradation of NRF-2. Glutamine, being an essential source of energy, Telaglenastat, Sapanisertib with good systemic tolerance, appear promising.	Preclinical
	Cervix: TCGA 7% - BioRAIDs 7%		MSU38225
	Ovary: TCGA 4%		Brusatol
	Endometrium: TCGA 9%		Early, combination trials
			Glutaminase inhibitors:
			Telaglenastat: 16 CT
			Sapanisertib: 24 CT
NOTCH1: GoF or LoF MM, TM, DD, SM		NOTCH1 displays cross-talks with many oncogenic signalling pathways, confers resistance to chemoradiation and has also been suggested as a biomarker for ICB.	Approved biomarker
	Cervix: TCGA 8% - BioRAIDs 7%	Antibody-based biologics: Demcizumab (NCT02259582), Dilpacimab, (targeting DLL4), Gamma-secretase inhibitors (Nirogacestat, AL101/BMS 90602) are under investigation.	ICB for notch1
	Ovary: TCGA 4%	Rova-T targeting DLL3 with payload was substandard in a phase 3 trial. A phase 3 trial with Nirogestat is not yet included at the time of submission.	Preclinical, early clinical
	Endometrium: TCGA 12%		Demcizumab: 6 CT
			Nirogacestat: 5 CT
			AL101: 4 CT
			Phase 3
			Nirogacestat: 1 CT
			Rova\xe2\x80\x91T: 1 CT

A, amplification; ARID1A, AT rich interactive domain 1A; CREBBP (CBP), cAMP response element-binding-protein; CC, cervical cancer; CSMD3, Cub and Sushi multiple domains 3; CT, computed tomography; DD, deep deletion; FDA, Food and Drug Administration; GoF, gain of function; GST, glutathione S-transferase; HDAC, histone deacetylase; ICB, immune checkpoint blockade; ICG-001, indocyanine green-001; IF-M, in-frame mutation; KMT2D, lysine methyltransferase 2D; LoF, loss of function; MCL, myeloid cell leukemia; MLL2, mixed lineage leukemia 2; MM, missense mutation; NFE2L2 (NRF-2), nuclear factor erythroid 2 related factor 2; NOTCH1, notch receptor 1; PIK3CA, phosphatidylinositol-4,5 bisphosphate 3-kinase catalytic subunit alpha; PTEN, phosphatase and tensin homolog; Rova-T, rovalpituzumab tesirine; SM, splice mutation; TCGA, The Cancer Genome Atlas; TM, truncating mutation; TP53, tumor protein 53; TR, translocation; UBR5, ubiquitin protein ligase E3 component n-recognin 5.

• Immune checkpoint inhibition has been suggested for CSMD3 mutated high grade serous ovarian carcinoma (HGSOC) since a sustained response to Pembrolizumab in patients without prior chemotherapy could be documented in HGSOC [6].

Ubiquitin protein ligase E3 component n-recognin 5 (UBR5) (alias EDD1) (LoF)

UBR5 encodes a progestin-induced protein and is a key regulator of the ubiquitin-proteasome system, a regulator of proteostasis and cell signaling, which when disrupted can affect DNA-damage response and apoptosis. In ovarian cancer (OC), LoF of UBR5 will prevent deactivation of Wnt/β-cat signalling. In CC, UBR5 is used by the HPV E6 oncogene to destabilize the tumor suppressor TIP60 through the proteasome pathway [15]. UBR5 LoF has been linked by several authors to Cisplatin resistance. Three proteasome inhibitors: Bortezomib, Carfilzomib, and Ixazomib have received regulatory approval in mantle cell lymphoma and multiple myeloma (MM). Two E3 ligase modulators, Lenalidomide and Pomalidomide, are used in the treatment of MM, and the Cereblon-E3-ligase modulator Iberdomide (CC-92480) is in development for relapsed and refractory MM. UBR5 alterations are found in 5% to 6% in CC.

AT-rich interaction domain 1A (ARID1A) (LoF)

ARID1A, a subunit of the switch/sucrose non-fermentable (SWI/SNF) complexes, facilitates the specific binding of this complex to chromatin, altering the accessibility of a variety of nuclear factors (NKs) to chromatin. It represses cell proliferation and prevents genomic instability by direct modulation of p53-mediated transcriptional regulation, as shown in OC. ARID1A LoF mutations or loss of protein expression have been linked to tumor progression and adverse prognosis in CC and, combined with PIK3CA mutations, to give a synergistic effect on tumor development in OC. ARID1A alterations are found in 7% to 9% in CC.

• Inactivation of the SWI/SNF complex is synthetically lethal with the inhibition of enhancer of zeste homologue 2 (EZH2). EZH2 inhibitor [8] GSK126, Tazemetostat, inhibited the catalytic subunit of the polycomb repressor complex 2, generating an epigenetic mark of a lysine 27 trimethylation on histone H3.

cAMP-response element binding protein (CREBBP) (short: CBP) (LoF)

CBP encodes a lysine acetyltransferase (KAT) with critical roles in embryonic development, growth control, and homeostasis. Functional p300/CBP complexes exert tumor-suppressive effects, promoting the functions of other tumor suppressors: p53, RB1, and BRCA1. Functional loss of CBP in SCLC reduced histone acetylation and transcription of cell adhesion genes, thereby leading to tumorigenesis. Similarly, CBP knockdown enhanced signaling of the RAS/RAF/MEK/ERK pathway [16]. CBP alterations are found in 8% to 9% in CC.

• Histone acetylation can be maintained by histone deacetylase (HDAC) inhibitors: Vorinostat, Pracinostat, and Panobinostat. Indocyanine green-001 (ICG-001) inhibits CBP binding to β-catenin, attenuating Wnt/β-cat signaling [10]. CBP/EP300 bromodomain inhibitors (BRDis) are under investigation.

Phosphatidyl-inositol 3 kinase carbonic anhydrase (PIK3CA) (activating mutations/gene amplification)

PI3K is a lipid kinase pathway, activated by cell-surface receptor tyrosine kinases such as Egf, Insulin, Igf-1, Vegf-A, and Pdgf receptors. The most important PI3K pathway proteins are those belonging to class IA (PI3KCA), composed of a catalytic subunit p110α and its associated regulatory subunit p85. PI3KCA catalyzes the formation of PIP3, a process that is reversed by the action of phosphatase and tensin homologue gene alteration (PTEN). Activating mutations of predominantly missense mutations, mapped to one of three hotspots in the p110α catalytic subunit of PIK3CA, have been identified in a wide range of tumors. Activation of mutations in PIK3CA in association with LoF alterations in epigenetic regulatory enzymes significantly impaired PFS in CC patients [3,4,17]. PIK3CA alterations are found in 39% to 40% in CC.

• PIK3CA inhibitor: Alpelisib showed prolonged PFS when associated with Fulvestrant in patients with PIK3CA-mutated, hormone receptor-positive, and HER2-negative advanced breast cancer (BC). Pan-PI3K inhibitors: Buparlisib and Copanlisib have shown objective response rates in hematological malignancies. The PI3KCA-AKT-mTOR pathway can also be targeted by mTor inhibitors (Everolimus) or AKT inhibitors (Capivasertib or Ipatasertib).

Tumor protein 53 (TP53) (LoF)

TP53 encodes a tumor suppressor protein (p53), inactivated in more than 50% of human cancers, arrests the cell cycle in response to a variety of cellular stresses, allowing DNA repair. P53 phosphorylation releases MDM2 binding and promotes p53 stabilization. In HPV-induced cancers, E6 viral proteins bind directly to p53, promoting its degradation via the ubiquitin pathway, inhibiting p53 related apoptosis. TP53 alterations are found in 7% to 9% in CC.

• APR-246 (PRIMA-1, MIRA-1) [12] binds covalently to the core domain of p53 protein, allowing the reactivation of the TP53 mutant. Glutathione S-transferase fusion proteins have shown preclinical efficacy, counteracting E6 disruption of p53. Inhibitors of P53-MDM2 interaction Milademetan, MDM2 inhibitors [18] are in early clinical development. Nutlins bind to the p53 domain of MDM2, stabilizing p53 expression.

PTEN (LoF)

PTEN is a TSG encoding a 3-phosphatase, which negatively regulates intracellular levels of phosphatidylinositol-3,4,5-trisphosphate, inhibits AKT/PKB and cell cycle progression. PTEN mutations occur at a significant rate in many human tumor types. Bi-allelic genetic loss triggers PTEN-induced cellular senescence (PICS). In contrast, a partial loss of PTEN, as in PTEN (+/−) cells, will not be sufficient to counter elevated PI3-kinase activities in cancers, but at the same time will not trigger PICS cellular senescence and death. PTEN LoF with additional loss of Rb1 has been shown to facilitate lineage plasticity and metastasis. PTEN alterations are found in 9% to 13% in CC.

• Downstream effectors such as AKT inhibitors: Ipatasertib or Capivasertib have been suggested.

Lysine methyl transferase 2D (KMT2D) (alias mixed leukemia lymphoma 2, MLL2 (LoF)

KMT2D functions as a major enhancer/regulator in cell development, in differentiation, and in tumor suppression. It is a component of a protein complex named ASCOM and encodes a protein that methylates histone 3 at Lys-4 to promote genome accessibility and transcription. It facilitates binding of CBP and/or Ep300 acetyl transferase. KMT2D mutations give rise to significant genomic instability in vivo which is partly explained by the fact that the ASCOM complex loses the ability to function as a coactivator of the tumor suppressor p53. LoF alterations of KMT2D have been found to be associated with poor PFS after chemoradiation [4] in all International Federation of Gynecology and Obstetrics stages of CC. KMT2D alterations are found in 15% to 16% in CC.

• BRDis halt the transcriptional cascade of oncogenes, with some showing clinically significant results and safety profiles of Birabrezib [19].

Nuclear factor erythroid 2 related factor 2 (NFE2L2) (alias NRF-2) (LoF)

NFE2L2 encodes a transcription factor, which is a member of a small family of basic leucine zipper proteins, which regulate genes containing antioxidant response elements in their promoters. Under normal, non-stressed circumstances, low cellular concentrations of the protein are maintained by proteasomal degradation. In cancer, the altered gene encodes NRF-2 protein isoforms lacking the KEAP1 interaction domain, resulting in NRF-2 stabilization, induction of transcriptional response, and NRF-2 pathway dependency. NFE2L2 alterations are found in 7% in CC.

• In preclinical studies, the small molecule ML385 bound to NRF-2 and inhibited downstream target gene expression [20]. Brusatol, a quassinoid from Brucea sumatrana seeds, equally exhibited tumor suppression.

Notch receptor 1 (NOTCH1)

NOTCH1 was thought to function as an oncogene or a tumor suppressor in different cancers and in different cell populations within the same tumor. It displays crosstalk with many oncogenic signaling pathways and with the tumor micro-environment. It can lead to acquired resistance to targeted therapies and standard chemoradiation [4]. It activates the transcription of target genes including HES1, CYCLIN D1, and c-MYC. Interestingly, in virus-induced cancer, increased Notch1 signaling has been suggested to down-modulate viral E6/E7 gene transcription. NOTCH1 alterations are found in 7% to 8% in CC.

• A Phase III trial of a NOTCH γ-secretase inhibitor, Nirogacestat, is tested in inoperable desmoid tumors (NCT03785964). Blocking antibodies against NOTCH ligands or receptors are in clinical testing: Demcizumab, an humanized immunoglobulin G (IgG)2 monoclonal antibody against DLL4, in combination with Pemetrexed and Carboplatin in metastatic NSCLC [21].

2. Protein overexpression or post translational modifications linked to poor outcome

β-catenin-phosphorylated in Serine 675 (Pβ-cat-Ser675)

Wnt/β-catenin signaling plays an important role in embryonic development and tissue homeostasis, promoting both cell proliferation and epithelial mesenchymal transition (EMT) (Table 2) [22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40]. The physiological binding of a Wnt ligand to cell surface receptors (Frizzled or LRP5-6), leads to inactivation of the β-catenin destruction complex. Therefore, normal β-catenin translocates into the nucleus and interacts with members of the DNA-binding T cell factor (TCF)/lymphoid enhancer factor family to activate Wnt target genes. In cancer, activating mutations in exon 3 of the β-catenin gene (CTNNB1), common in endometrial cancer, are rare in CC. In a CoxBoost analysis of the clinical BioRAIDs CC dataset, post-translational phosphorylation patterns of protein β-catenin were highly significant. While Pβ-cat-Ser552 was associated with excellent outcomes after standard therapy, Pβ-cat-Ser675 was highest on the list of markers associated with poor outcome [3].

Table 2. Therapeutic interventions targeting PPI.

Quantitative (above median) or qualitative alteration in tumor	Physiopathology - role in treatment resistance	Alternate treatments
Pβ-cat-Ser675 (post-translationally modified form of β-catenin, phosphorylated in serine 675)	Phosphorylation at the Ser675 site promotes binding of β-catenin to CBP, leading to maintenance of potency and self-renewal of stem/progenitor cells, while binding of β-catenin to p300 coactivator leads to activation of genes involved in differentiation.	Preclinical and early clinical
	ICG-001 [22,23] is a CBP/catenin antagonist, which binds specifically to CBP, but not to the histone acetylase p300.	ICG-001
		The ICG-001 derivative PRI-724 is in clinical exploration
		HDAC inhibition
Protein 14-3-3 β	The 14-3-3 family of adapter proteins, with several hundred protein interaction partners, is involved in key cellular processes: Raf kinases, cell cycle phosphatase Cdc25, transcriptional modulator YAP and tumor suppressor p53. The β isoform is elevated in cancer patients.	Preclinical
		Targeting strategies reviewed in [24]
Phospho Eef2K-Ser366 (eukaryotic elongation factor 2 kinase)	Facilitates protein synthesis. High levels of this kinase allow high Eef2 activity, facilitating protein synthesis. Eef2K is activated under poor nutrient conditions, protects against nutrient deprivation.	Preclinical
		Reviewed by Liu et Proud 2016 [25], Zhang et al. [26]
Hsf1	Hsf1, phosphorylated at Ser326 by oncoproteins, binds to heat shock promoter elements, promotes cell migration and EMT [31].	Preclinical and early clinical
	Hsf1-pSer326 predicts response to bortezomib-containing chemotherapy in pediatric AML: a COG report [27].	Bortezomib predicted
		NPX800: NCT05226507
		Inhibitors reviewed by Chatterjee and Burns [28]
Tnrfsf18/Gitr (alias GITR)	Immune modulator by inhibition of Treg and activation of CD8+ T effector cells. The anti-Pd-1-Gitr-L bispecific AB is a construct that has demonstrated dose-dependent and immunological tumor growth inhibition in syngeneic, genetically engineered, and xenograft humanized mouse tumor models [32].	Phase ½ and combination trials
	The 8/13 patients (ICB)-naïve melanoma patients receiving MK-4166 in combination with ICB achieved a clinical response [33].	INCAGN01876: 7 CT
		BMS-986156: 2 CT
		REGN6569: 1 CT
		GWN323: 1 CT
		OMP-336B11: 1 CT
		ASP1951: 1 CT
		MK-4166: NCT02132754
Glut1	Facilitates glucose transport with implications for cancer prognosis. Glut1 inhibition by BAY-876 specifically blocked the growth of RB1-positive cancer in preclinical assays.	Preclinical
		BAY-876
Pkm2	Enzyme regulating the metabolism of tumor cells, promoting chemoresistance.	Preclinical
	Nuclear transcription cofactor following acetylation of Pkm2 by p300 acetyltransferase at K433 with subsequent binding to β-catenin p300, cyclin D1, HDAC3. NPD10084 interrupted PPI between Pkm2 and β-catenin, suppressing downstream signaling.	NPD10084
		Clinical
		Pkm2 detection in cancer NCT03539731
		Benserazide [29] constituent of Modopar, withdrawn
Pkcδ	Many PkC inhibitors have entered clinical trials with limited success. A Phase III trial with the selective PkCβ inhibitor, Enzastaurin was negative.	Preclinical
	Recent studies favor PkC function as tumor suppressors, suggesting that the focus should be on restoring rather than inhibiting PkC function [34].	PKC inhibitors
		Clinical phase 3
		Enzastaurin
Ca IX	Ca IX maintains alkaline intracellular pH and exacerbates extracellular acidosis [35] Ca IX inhibitors are emerging agents for the treatment and imaging of hypoxic tumors. Most small molecule Ca IX inhibitors such as Belzutifan (also an oral HIF-2α inhibitor) appeared safe and showed activity. Girentuximab is a radiolabeled chimeric antibody for detection of Ca IX positive cancers.	Clinical
		Belzutifan: NCT03401788
		SLC-0111/OZM-055
		MK-6482-004: NCT04976634
		PT2385: NCT03216499
		Girentuximab: 7 CT [30]
Vimentin	Cytoskeletal protein playing an essential role in EMT by increasing the migratory/invasive capacity of cancer cells in interaction with Erk, AKT1, and 14-3-3 protein. Monoclonal antibody 86C increased sensitivity to temozolomide in glioma stem cells. Silibinin A: active constituent of silymarin downregulates vimentin.	Preclinical
		Monoclonal antibody 86C [37]
		Silibinin A
Nbs1	Nbs1 protein interacts with Atm/Atr-dependent checkpoint.	Preclinical and early clinical
	Early phase trials by inhibitors of ATR, CHK1 and WEE show signs of activity.	BAY 1895344 (Elimusertib)
		Reviewed in [38]
c-Abl	Regulates proliferation, survival, adhesion, and apoptosis. Activated by Egfr, Igf-1r and Src, c-Abl phosphorylates Plk1, thereby inhibiting Plk1 degradation, leading to tumor progression [39]. Numerous Abl kinase inhibitors (n=136) in development.	Preclinical and clinical
		Imatinib mesylate
		Nilotinib
		Dasatinib
		Bosutinib
		GNF-2, GNF-5, ABL-001
Phospho-Pten-Ser380-Thr382-383 (inactive phosphor protein form)	Active PTEN acts as lipid phosphatase and tumor suppressor to convert PIP3 to PIP2. Its activity is regulated by phosphorylation on a cluster of four Ser and Thr residues near the C-terminus by multiple kinases, including casein kinase 2 and GSK3β.	Treatment by phosphatases was suggested to reactivate this inactive form of Pten
Phospho AKT Ser473 (alias protein kinase B)	Serine/threonine kinase involved in proliferation, migration, anti-apoptosis, and maintenance of metabolic homeostasis and immunomodulation.	CT
		Pan-AKT kinase inhibitors: reviewed in [36]
Egfr	Egfr triggers many pathways and regulates cancer stem cells. Reviewed in [40]. The anti-apoptotic Bax/Bcl-2 cascade has been recently shown to be regulated by the EGFR pathway. Venetoclax is an FDA approved drug in combination with hypomethylating agents for unfit AML patients. Venetoclax is not active if MCL-1 is high.	Combination trials
		A: Anti-Egfr AB
		B: EGFR tyrosine kinase inhibitors
		C: Bax/Bcl-2 inhibitors
		ABT-199 (Venetoclax): 381 CT

AML, acute myeloid leukemia; Ca IX, carbonic anhydrase 9; c-Abl, abelson tyrosine kinase 1; CBP, CREBBP, cAMP response element-binding-protein; COG, Children's Oncology Group; CT, computed tomography; Egfr, epidermal growth factor receptor; EMT, epithelial mesenchymal transition; Glut1, glucose transporter 1; HDAC, histone deacetylase; Hsf1, heat-shock factor 1; ICB, immune checkpoint blockade; ICG-001, indocyanine green-001; Nbs1, nijmegen breakage syndrome 1; Pβ-cat-Ser675: β-catenin-phosphorylated in serine 675; Pkcδ, protein kinase C delta; Pkm2, pyruvate kinase iso-enzyme 2; PPI, protein-protein interaction; PTEN, phosphatase and tensin homolog; Tnrfsf18, tumor necrosis factor receptor superfamily member 18.

• Indirect targeting of Pβ-cat-Ser675 via interference of CBP binding to β-catenin by ICG-001, has recently been suggested in CC [3]. ICG-001 enhanced drug sensitivity in acute lymphoblastic leukemia and nasopharyngeal carcinoma [3], suppressing cell line growth, stem cell-like properties, causing chemoresistance [23]. Small molecules that bind directly to β-catenin have been identified.

Protein 14-3-3 β

Protein 14-3-3 β is a member of phosphorylated serine/threonine proteins that bind to a variety of kinases, phosphatases, transmembrane receptors, and transcription factors, acting as adaptors and displaying chaperone-like activity. More than one hundred small molecules are thought to interact with 14-3-3 β in a phosphorylation-dependent manner, affecting both AKT and apoptosis signaling. In cancer, they activate tyrosine 3-monooxygenase or tryptophan 5-monooxygenase. Elevated cytoplasmic protein expression or high transcriptional levels of 14-3-3 β were associated with advanced disease and aggressive tumor characteristics in vulvar cancers [41], BC, hepatocellular carcinoma (HCC), and osteosarcoma (OS) cells. Specific isoforms were correlated with the infiltration of specific inflammatory cell infiltrates. Recent developments in modulators of 14-3-3 protein-protein interactions (PPIs) have been reviewed by Stevers et al. [24].

• Based on functional activity of protein 14-3-3, both methylation inhibitors and HDAC inhibitors may be candidate cancer therapies.

Phospho eef2k-Ser366 (eukaryotic elongation factor 2 kinase, eef2k; inactive hosphor protein form)

Functional eef2k regulates the elongation stage of protein synthesis by phosphorylation leading to inhibition of its substrate, eef2k [42]. High levels of eef2K, when phosphorylated in Ser366, are functionally inactive, allowing high levels of eef2k levels leading to enhanced protein synthesis. Eef2k-Ser366 also promotes glycolysis via activation of pkm2 transcription. Activated mTorc1 inhibits eef2k function by phosphorylation at residues: Ser70, Ser78, Ser359, Ser366, Ser392, Ser396, and Ser470. Activation of AMPK can in turn induce phosphorylation at sites: Ser398 or Ser491, thereby activating eef2k leading to inhibition of its substrate.

• Tumor cells expressing high levels of the inactive eef2k-Ser366 may benefit from treatment with Bortezomib (Velcade) or Rapamycin (Everolimus). Brain derived neurotrophic factor induced phosphorylation of eef2k in Ser366, decreasing its kinase activity. This action was inhibited by rapamycin. Pharmacological activation of eef2k by Nelfinavir downregulated eef2k and inhibited cancer progression [43]. Resveratrol activated AMPK, resulting in phosphorylation of residue Ser398 and activation of eef2k, thereby inhibiting eef2k activity. Fluoxetine, a selective serotonin reuptake inhibitor, the clinical anti-depressant “Prozac”, induced autophagic cell death in triple-negative breast cancer (TNBC) cells. Rottlerin significantly inhibited eef2 and proliferation in human glioma cells.

Heat-shock transcription factor 1 (Hsf1)

Hsf1 is a master regulator of heat shock response. It trimerizes after phosphorylation and translocates to the nucleus where it binds to DNA heat-shock elements to trigger the recruitment of a pre-initiation complexes comprising RNA polymerase II and transcription factors. High-level expression of Hsf1 and nuclear localization have been reported in carcinomas of the cervix, colon, lung, pancreas, and prostate and were strongly associated with reduced survival in colon and lung cancers. HSF1 was identified as one of only six potent metastasis-promoting-genes in a genome-wide screen of malignant melanoma cells. Under non-stressed conditions, inactive monomeric Hsf1 are repressed by products of their own transcriptional target, heat shock protein.

• The Hsf1 inhibitor direct targeted HSF1 inhibitor physically engages Hsf1 and selectively causes nuclear Hsf degradation and acts as a potent tumor growth suppressor [44].

Tnfr superfamily member 18; glucocorticoid-induced TNFR-related protein (Tnfrsf18) (alias Gitr)

Tnfrsf18, alias Gitr, is a cell surface receptor constitutively expressed at high levels on T regulatory cells (Tregs) and at low levels on naïve and memory T cells. Modulation of the Gitr/Gitr ligand axis suggested that this pathway could both inhibit Treg function and activate CD8+ T effector cells by upregulating CD25, inducing interleukin-2 and interferon-γ expression. In mature T-regs, FoxP3 promotes high-level Gitr expression. In conventional T cells, GITR signaling is mediated by activation of NF-κB and MAPK pathways.

• GITR agonist antibodies [45] or recombinant Gitr ligands showed antitumor activity [46]. A first in-human phase 1 trial of Gitr agonism with the anti-Gitr antibody TRX518 (NCT01239134) was safe and reduced circulating and intratumoral Treg cells. No substantial clinical responses has been observed [46], but combination trials are ongoing.

Glucose transporter 1 (Glut1)

The Glut family is composed of 14 members of a family of hydrophobic proteins, expressed in the basal and immediate supra-basal layers of the epidermis. GLUT1 transcription is upregulated by PIK3/AKT signaling while p53 regulates glycolysis by inhibiting Glut1 and Glut4 expression. LoF mutations in p53 lead to upregulation of glucose transporters, enhancing the influx of glucose required for energy for cancer cell growth. Many tumorigenic cells are characterized by high catabolic utilization of glucose and an increased number of specific glucose transporters. Human CC selectively overexpresses Glut1 and sufficient intracellular glucose supply is a prerequisite for enhanced HPV transcription and expression of oncoproteins E6 and E7 [47]. Glut1 inhibition blocked cell growth in Rb1-positive TNBC cells.

• STF-31, a pyridyl anilino thiazole, selectively kills renal cancer clear cells by specifically targeting glucose uptake through Glut1 and the ho pathway.

Pyruvate kinase-isozyme type M2 (Pkm2)

Pkm2 is a rate-limiting glycolytic enzyme, catalyzing the irreversible transphosphorylation producing pyruvate and ATP. Pkm2, one of four isoforms, is expressed in tissues with anabolic functions and tumor types such as gastrointestinal, HCC, and NSCLC. Insulin upregulates Pkm2 expression. Nuclear Pkm2-dependent β-catenin transactivation is required for epithelial growth factor receptor (Egfr)-promoted tumor cell proliferation [48]. This results in removal of HDAC3 from the cyclin D1 promoter, histone H3 acetylation, leading to cyclin D1 and c-Myc expression. Overexpression of metabolic markers PKM2 and Ldh5 correlated with aggressive clinicopathological features and adverse patient prognosis in tongue cancer, while knockdown of PKM2 suppressed tumor progression in CC cells by modulating EMT via Wnt/β-catenin signaling [49].

• NPD10084 interrupted signaling between Pkm2 and β-catenin or Stat3, and suppressed downstream signaling. PKM2 activation could be prevented by acetylation via functional p300 acetyltransferase at lysine-433 (Research 2013). Benserazide, used in Parkinson disease, had an anti-Pkm2 activity in melanoma cells [29].

Protein kinase C-subunit α (Pkcδ)

Pkcδ belongs to a family of serine/threonine protein kinases and is activated by diacylglycerol. There is conflicting evidence as to whether Pkcδ act as oncogenes or as tumor suppressors, as overexpression and loss of Pkcδ in colon cancer cells has been reported to decrease tumorigenicity. Precise phosphorylation sites are likely important for activation or suppression of function. B-cell deficiency and severe auto-immunity were related to Pkcδ deficiency.

• Combined inhibition of Pkcδ and Egfr was necessary to induce marked regression of resistant NSCLC tumor cells with EGFR mutations [50]. Rottlerin has been suggested as a selective Pkcδ inhibitor.

Carbonic anhydrase 9 (Ca IX)

Ca IX is a tumor-associated cell-surface glycoprotein induced by hypoxia. It is essential for acid-base balance, maintaining alkaline intracellular pH, critical for proliferation of cancer cells in hypoxic conditions. It can destabilize intercellular adhesion contacts through competitive binding to β-catenin. It exacerbates extracellular acidosis, thereby activating proteases to cleave extracellular matrix, facilitating EMT and invasion, and supporting inflammation and angiogenesis. Hif and Sp1 transcription factors cooperate towards induction of Ca IX in conditions of increased cell density and acidosis. Transcription of the CA IX gene increases in response to oncogenic signaling in MAPK and PI3K pathways.

• A phase 1 study of SLC-0111 reported stable disease >24 weeks in 2 patients and a combined study with Gemcitabine for metastatic pancreatic ductal cancer (Ca IX positive tumors) is ongoing: NCT03450018. Girentuximab, an anti Ca IX IgG1 κ light chain chimeric version, has not shown clinical benefit as single drug in renal cell carcinoma (RCC). DH348, a bioreductive nitroimidazole-based anti-Ca IX drug, reduced tumor growth in mice and sensitized tumors to irradiation in a Ca IX–dependent manner [51].

Vimentin

Vimentin is one of the most widely expressed and highly conserved proteins of the type III intermediate filaments protein family which, along with microtubules and actin microfilaments, make up the cytoskeleton. Vimentin regulates 14-3-3 complexes by depleting the availability of free protein 14-3-3 controlling various intracellular signaling and cell cycle pathways. Vimentin overexpression is frequently associated with increased migratory/invasive ability of cancer cells and generally considered an indicator of stem cell activity and poor prognosis. In the cytosol, vimentin has been shown to interact with phosphorylated Erk, to be phosphorylated by Akt, thereby preventing caspase-induced proteolysis, leading to freely available vimentin, able to participate in the increase of the migratory and invasive capacity of cells.

• Silibinin A: the active constituent of silymarin, with chemo preventive and anticancer activities, has been shown to inhibit the invasion, motility, and migration of prostate cancer cells via downregulation of vimentin and matrix metalloproteinase-2.

Nijmegen breakage syndrome 1 (Nbs1)

Nbs1 is a recessive genetic disorder due to a defective response to face DNA damage, characterized by immunodeficiency and a high frequency of malignancies. It is a member of the MRE11-RAD50-NBS1 (MRN) double-strand break repair complex, a sensor to DNA damage, telomere dysfunction and viral invasion. Overexpression of Nbs1 has been shown to induce EMT and is correlated with poor response and shorter OS after chemotherapy and radiotherapy. Co-expression of Nbs1 and Snail predicted head and neck squamous cell carcinoma (HNSCC) metastasis.

• Disruption of MRN function by silencing Rad50 or Nbs1 has been suggested to sensitize cells to cisplatin, radiotherapy, or other DNA damaging agents. Protein trafficking of MRN components along microtubules can be blocked by microtubule-targeting agents. Mirin, an inhibitor of the MRN complex, inhibits MRE11-associated exonuclease activity, prevents MRN-dependent ataxia, telangiectasia mutated (ATM) activation, and abolishes repair in mammalian cells [52]. The HDAC inhibitor Panobinostat has been shown to downregulate MRE11 and to radiosensitize tumor cells [53].

Abelson murine leukemia viral oncogene homolog 1 (c-Abl)

Abelson tyrosine-protein kinase 1 is a member of the Src family, containing: Src homology 2 (SH2) and 3 (SH3) which regulate the activity of c-Abl protein, a long C-terminal tail which contains a DNA-binding and an actin binding domain. c-Abl in conjunction with (polo-like-kinase-1 (Plk1) were shown to be overexpressed in CC. Phosphorylation of Plk1 by c-Abl inhibited the ubiquitination and degradation of Plk1, enhancing its activity, leading to cell-cycle progression and tumor growth. Several models for G1 cell cycle arrest by c-Abl have been proposed, involving actions on p53, MDM2, and Rb.

• Imatinib inhibits Abl kinase activity. Nilotinib and Dasatinib together inhibit all common Imatinib resistant mutations. Ponatinib has shown promising results in patients with the T315I mutation or with prior resistance to Nilotinib/Dasatinib [54]. A combination treatment of GNF-2 in association with Nilotinib has been shown to prolong survival in a Bcr-Abl T315I mouse model.

Phospho-Pten-Ser380-Thr382-383 (inactive phospho protein form)

LoF of PTEN is one of the most frequent genetic alteration in human tumors and functional alterations of the PTEN protein are governed by phosphorylation on a cluster of four Ser and Thr residues near the C terminal tail. These post-translational modifications modulate protein structure, activity, location, and interactions [55]. While cytoplasmic PTEN is primarily involved in the regulation of PI3K/PIP3 signaling, nuclear PTEN exhibits phosphatase-independent tumor suppressive functions. Nuclear PTEN binds to the anaphase-promoting complex and complexes with p300/CBP acetyltransferase to promote p53 acetylation in response to DNA damage. P53 acetylation in turn enhances PTEN-p53 interaction and is crucial for the induction of Rad51 which regulates DNA double-strand break repair [36].

• In light of the high protein levels of inactive forms Phospho-Pten-Ser380-Thr382-383 in cancer, Henager et al. [56] suggested using alkaline phosphatase to achieve rapid dephosphorylation, enabling rapid activation of PTEN, relevant for cellular phospholipid signaling.

Phospho-Akt Ser473 (active phospho protein form)

Akt regulates cell survival, proliferation, apoptosis, and glycogen metabolism. Akt overactivation is a common molecular characteristic of human malignancies, subsequent to Erb-B2 amplification, EGFR/PI3K mutation, PTEN LoF, or direct mutations or gene amplifications of AKT. The Akt protein is activated by phosphorylation on Thr308 or Ser473 and in turn phosphorylates a variety of downstream substrates including Gsk3β and Mdm2. Ser473 post translational modified form of Akt was a significant predictor of poor outcome and increased PSA levels in prostate cancer with a ROC curve of 0.84 for the combination of high Akt Ser473 and low pErk.

• Akt inhibitors in combination treatments have been shown to be clinically effective: reviewed by Song et al. [57]. Capivasertib + Paclitaxel as first-line treatment for TNBC resulted in significantly longer PFS and OS. GDC-0068 (Ipatasertib), a pan-Akt inhibitor, showed longer PFS compared in combination with Paclitaxel in TNBC.

Egfr

Egfr belongs to the ErbB family of cell surface protein receptors known to modulate growth, adhesion, migration, and survival of cancer cells. Overexpression of Egfr is frequent in CC and gene amplification is present in 10%–20%. Aberrant Egfr activity triggers the KRAS-BRAF-MEK-ERK, PI3K,αgamma, Akt, and STAT signaling pathways.

Blockade is effective only in the absence of additional downstream activations.

• A: anti-EGFR monoclonal antibodies Cetuximab: bind to the second (L2) domain of EGFR thereby blocking its downstream signaling by prompting internalization of the receptor and interfering with ligand-receptor interaction [58]. Panitumab is approved for RAS wild type colorectal cancers [59].

• B: Numerous tyrosine kinase inhibitors are in clinical use and in development: including Gefitinib, Erlotinib, Dacomitinib, Afatinib, Osimertinib, Vandetanib, Lapatinib, and Canertinib.

3. Genetic alterations linked to good outcomes following chemoradiation, suggesting therapeutic exploitation for synthetic lethality

Alpha thalassemia/mental retardation syndrome, X-linked (ATRX) (LoF)

The ATRX protein product orchestrates the deposition of the histone H3.3 variant in telomeres and in pericentric DNA, thereby maintaining genomic stability in these repetitive regions (Table 3) [60,61,62]. ATRX belongs to the SWI/SNF gene family. It contains an ATPase-helicase domain and has been implicated in the regulation of DNA damage repair, both through non-homologous end-joining (NHEJ) and homologous recombination repair. In cancer, ATRX LoF may occur though missense mutations, deletions, or gene fusions. It frequently correlates with other molecular changes such as the alternative lengthening of telomere phenotype, which includes platelet-derived growth factor receptor alpha gene amplification and TP53 mutations. Cells derived from an ATRX LoF mouse glioma model were defective in NHEJ and more susceptible to DNA damaging treatments [63]. ATRX knock-out leads to failure in triggering Atm phosphorylation. ATRX LoF in patient tumors leads to specific DNA damage repair defects that can be exploited therapeutically [3,4,60]. ATRX alterations are found in 6% to 9% in CC.

Table 3. Exploring treatment de-escalation for genetic alterations associated with good outcome.

Gene name, alteration type and frequency		Role in cancer: genetic alteration which might be therapeutically exploited	Potential treatment de-escalation
ATRX: LoF by A, MM, TM, SM		ATRX, being a critical regulator in promoting the transition of quiescent cells to therapy-induced senescence, its LoF resulted in specific DNA damage repair defects that can be exploited therapeutically [60].	Multiple Parp inhibitors, Atm inhibitors, wee1 kinase inhibition, DNA damaging agents as well as sapacitabine and irinotecan have been suggested for ATRX LoF alterations.
	Cervix: TCGA 6% - BioRAIDs 9%
	Ovary: TCGA 4%
	Endometrium: TCGA 15%
MED13 (alias thyroid hormone receptor-associated protein): LoF by A, TM, MM, F		MED13 LoF may increase susceptibility to cancer cell damage by chemo-radiation, possibly acting via perturbation of Cdk8 function involved in β- catenin signaling.	Cdk8/19 inhibitor Senexin
	Cervix: TCGA 4% - BioRAIDs 5%
	Ovary: TCGA 4%
	Endometrium: TCGA 15%
KMT2C (alias MLL3): LoF by DD, TM, MM		KMT2C is a histone methyltransferase that methylates ‘Lys-4’ of histone H3, a specific tag for epigenetic transcriptional activation. Its function is necessary for chromatin modifications and epigenetic gene regulation.	ICB [61]
	Cervix: TCGA 20% - BioRAIDs 15%
	Ovary: TCGA 13%
	Endometrium: TCGA 21%
CASP8: LoF by DD, TM, MM		Conversely, functional retention of CASP8 in glioblastoma was thought to interfere with sensitivity to radio and chemotherapeutic approaches by improving DNA damage repair and activating NF-κB and cytokine production [62].	ICB [61]
	Cervix: TCGA 7% - BioRAIDs 7%		LoF of CASP8 may enhance the sensitivity of cancer cells to DNA damaging agents, independent of apoptosis
	Ovary: TCGA 2,5%
	Endometrium: TCGA 21%

A, amplification; ATRX, α-thalassemia/mental retardation, X-linked; CASP8, caspase-8; DD, deep deletion; F, fusion; ICB, immune checkpoint blockade; KMT2C, lysine methyl-transferase 2C; LoF, loss of function; MLL3, mixed leukemia lymphoma 3; MM, missense mutation; NK-κB, nuclear factor kappa B; SM, splice mutation; TCGA, The Cancer Genome Atlas; TM, truncating mutation.

• Significant sensitivity has been demonstrated to multiple Parp inhibitors [60] (Talazoparib), the Atm inhibitor KU60019, and DNA damaging agents Sapacitabine and Irinotecan. The Olaparib/Irinotecan in combination was effective in in vivo models of neuroblastoma. Mice harboring ATRX-deficient tumors were highly sensitive to ICB. In a synthetic lethal screen using CRISPR-Cas9 genome editing; the kinase Wee1 was one of 58 genes, only required for the cell growth of ATRX null cells, suggesting that Wee1 kinase inhibitors are potentially synergistic drugs in tumors with LoF of ATRX [64] such as AZD1775 (Adavosertib).

MED13 (alias thyroid hormone receptor-associated protein) (LoF)

MED13 in association with MED12, CDK8, and CYCLIN C constitutes a four-subunit “kinase” module that exists in variable association with a 26-subunit Mediator core involved in developmental and oncogenic signaling, conveying information from regulatory proteins to RNA polymerase II transcription. Mediator is recruited to promoters by transcriptional activators, by estrogen or androgen nuclear receptors. Mediator-associated cdk8 kinase activity in association with β-catenin-regulated target gene transcription was required for cell transformation and proliferation of colon cancer cell [65]. It has been suggested that mutations would affect the expected transcriptional activation by MED13 on Mediator, resulting in reduced mitochondrial biogenesis and fatty acid oxidation. Conversely, in BC, MED13 gene amplification is common (9.7%) and, together with cdK8 protein overexpression, was associated with shorter recurrence-free survival. MED13 alterations are found in 4% to 5% in CC.

• The increased sensitivity to DNA repair inhibition in the presence of MED13 LoF can be exploited therapeutically.

Lysine methyl transferase 2C (KMT2C) (alias mixed leukemia lymphoma 3, MLL3) (LoF)

KMT2C is a histone methyltransferase that methylates ‘Lys-4’ of histone H3, a specific tag for epigenetic transcriptional activation. Its function is necessary for chromatin modifications and epigenetic gene regulation. In cancer, a partial duplication of KMT2C has been found in the juxta-centromeric region of several chromosomes. The genes encoding KMT2C [66] and/or KMT2D are subject to deletion, mutation or partial duplication in many cancer types. The KMT2C gene is located in a critical region of chromosome 7 and commonly deleted in MM. In pancreatic ductal adenocarcinoma, similar to CC, a group of patients with KMT2C LoF has been identified with a more favorable prognosis after treatment impacting the DNA repair axis. In endometrial cancer, the mutational status of these genes predicted the degree of myometrial invasion, a critical prognostic feature. Although there is evidence suggesting that KMT2C LoF may render cancer cells more susceptible to chemotherapy [4], conversely, it has been associated with hormonal resistance in BC cell lines. Immunotherapy may be worth evaluating, as LoF alterations of KMT2C have been shown to be significant predictors of ICB response in cohorts [61]. KMT2C alterations are found in 15% to 20% in CC.

Caspase 8 (CASP8) (LoF)

Casp8 is the most upstream protease in the caspase activation cascade responsible for cell death mediated by Tnfrsf6/FAS and Tnfrsf1A via the death-inducing signaling complex. CASP8 contributes to DNA damage response functionality and was originally identified as an essential player in apoptosis. Under physiological conditions, Casp8 cleaves and activates a series of Caspases as well as Parp1. In HNSCC, the majority of CASP8 mutations were oral cavity tumors [67]. CASP-8 LoF was associated with good prognosis in HNSCC (The Cancer Genome Atlas dataset) similar to CC treated with chemoradiation [4]. CASP8 alterations are found in 7% in CC.

• Emerging corroborating evidence suggested that the functional retention of CASP8 in glioblastoma may interfere with the sensitivity to radio and chemotherapy through improved DNA repair [62].

4. Changes in protein expression/activation in association with good outcome

β-catenin protein phosphorylated in Serine 552 (Pβ-cat-Ser552)

Wnt/β-catenin signaling plays an important role in embryonic development, tissue homeostasis, and EMT (Table 4) [68,69,70,71,72,73,74,75,76,77,78,79,80]. Goretsky et al. [68] reported that β-catenin was active in a low molecular weight (Pβ-cat552) nuclear form, showing increased transcriptional activation, cell proliferation, and growth of tumor xenografts. Ser552 to alanine substitution abrogated these effects. Phosphorylation in Ser552 supposedly promotes stabilization of the protein, enhancing TCF-mediated transcription. When Wnt signaling is turned off, FBXW7 targets β-catenin for ubiquitin-mediated destruction in the cytoplasm, preventing nuclear entry. Several proteins (adenomatous polyposis coli, Chibby, and protein 14-3-3) facilitate the nuclear export of β-catenin [81].

Table 4. Exploring treatment de-escalation for protein forms associated with treatment response.

Quantitative/functional alteration	Role in cancer physiopathology	Potential treatment de-escalation
Pβ-cat-Ser552	Phosphorylation in Serine 552 by Akt leads to its dissociation from cell-cell contact and enhanced transcriptional activity [68].	Preclinical
	A disubstituted derivative of 8-methoxyquinazoline (18B) has recently been designed and shown to be able to disrupt the β-catenin/TCF4 signaling pathway, but the phospho type of β- catenin involved is unclear.	Metformin dramatically inhibited β-catenin mRNA and protein expression [73].
		HMQ-T-F5 [74] inhibited β-catenin production and nuclear import.
Ido	Ido was shown upregulated by CD8+ T cells. High Ido expression in the tumor microenvironment was thought to block vaccine efficacy. Ido inhibitors have been suggested useful as an adjunct to immunotherapy or vaccines, with little evidence of clinical efficacy so far.	Early clinical
		A recent phase III ECHO 301 trial testing the combination of Epacadostat with Pembrolizumab in melanoma did not show superior outcome compared to Pembrolizumab alone [75].
ATP5A (ATP synthase)	Involved in oxidative phosphorylation. Mutations in mitochondrial DNA, leading to defective protein synthesis and downregulation of up to 23/29 subunits of the Atp Synthase are common in cancer [69].	Translational research
		ATP synthase is a viable new drug target for the development of antimicrobials (Bedaquiline) approved in 2012 for the treatment of tuberculosis [76].
Jnk-Sapk1 (alias MAPK8)	Acts as tumor suppressor.	Preclinical
	High levels of stromal expressions have been associated with good prognosis in esophageal squamous cell carcinoma [77] and may be indicative of low PI3K-AKT pathway activity.	Glabridin, a flavonoid extracted from licorice [70] can lead to apoptosis through the JNK1/2 pathway.
44-42 Mapk1,2 (alias p44 Erk1/Erk2)	High levels of 44-42 Mapk1,2 can cause cell senescence by oxidative stress.
Dhfr	Involved in metabolism of nucleotides and amino acids.	Preclinical
	Elevated Dhfr levels are correlated with methotrexate resistance, leading to poor prognosis in many cancers [71] while in BioRAIDs elevated Dhfr levels were not an obstacle for response to chemoradiation.	Downmodulation of Golm1 [78] or the association of Silmitasertib (CX-4945) with a selective CK2 inhibitor may enhance the anti-tumor effect of methotrexate.
Gsk3 (isoforms alpha and β)	Negative regulator of glucose homeostasis, inhibitor of the Wnt/β-catenin pathway [72], induces the pro-aoptotic protein MCL-1.	Translational research
		Inactive or low expression of Gsk3β was associated with CC progression (p=0.024).
NdufA10	Ndufs interact with ATP5 gene products and are involved in mitochondrial energy production.
Gata3 (alias: trans-acting T-cell-specific transcription factor)	Role in maintaining the differentiation and adhesion of luminal epithelial cells, IHC staining reported in 20%–25%, often focal and weak [79], diagnostic for mesonephric type.	Preclinical
		Anti-adipogenic transcription factor
P27 (alias KIP1 or Cdk1b)	The tumor suppressor p27/kip1 (p27) inhibits cyclin/Cdk complexes and halts cell cycle progression.	Preclinical
		The BRD4 inhibitor IBET upregulates the stability of p27kip/cip protein in neuroendocrine tumor cells [80].
Ptp1B	Ptp1b has been reported to promote apoptosis through downregulation of pro-survival RTK signaling.

CC, cervical cancer; Cdk, cyclin dependent kinase; Dhfr, dihydrofolate reductase; Golm1, golgi membrane protein 1; Gsk3, glycogen synthase kinase 3; Ido, indoleamine 2,3 oxygenase; IHC, immunohistochemistry; NdufA10, NADH ubiquinone oxido-reductase subunit A10; Jnk-Sapk1, c-Jun N-terminal kinase stress-activated protein kinase; MAPK, mitogen-activated protein kinase; Pβ-cat-Ser552, β-catenin protein phosphorylated in Serine 552; Ptp1B, protein tyrosine phosphatase non-receptor type 1.

While proliferating tumor cells are more likely to be affected by chemotherapy than non-replicating stem cells, our results suggest that CC with high expression of the Pβ-cat-Ser552 isoform may be more likely to respond to chemoradiation [4], await confirmation.

Indoleamine 2,3-dioxygenase (Ido)

Ido catalyses the rate-limiting first step in the catabolism of the essential amino acid tryptophan along the kynurenine pathway, contributing to maintain immune homeostasis, preventing autoimmunity. Ido secreted by stem cells has been shown to exert immune regulatory functions in pregnancy. Controversial reports exist as to its relation to outcomes in tumors. A series of reports document high Ido expression in tumors in relation to tumor-associated inflammation, immune evasion [82], EMT [83], and poor outcome, but also with T cell infiltrates, programmed death-ligand 1 (PD-L1) expression and B-cell mediated autoimmunity. High Ido levels had been associated with resistance to ICB [84] while high levels in the tumor microenvironment appeared to be associated with good outcome in a large prospective study of chemotherapy-treated lymphomas [85] similarly to the prospective BioRAIDs [2] CC study with chemoradiation as the main treatment. Potentiation of vaccine-induced immunity by Ido inhibition has led to preclinical testing of Ido inhibitors. Further studies are needed to clarify the conditions under which Ido inhibitors may be of value in an anti-cancer strategy.

ATP5A (ATP synthase)

ATP5A, a mitochondrial membrane ATP synthase, is involved in oxidative phosphorylation. A hallmark of cancer is its extensive metabolic reprogramming, requiring large amounts of ATP. Mutations in mitochondrial DNA, resulting in deficient expression of ATP5A, have been associated with RCC [69] or HNSCC and confirmed at the protein level by Western Blot and immunohistochemistry. Dysregulation of ATP5A expression, activity or localization is linked to other human diseases, including diabetes and Alzheimer’s and Parkinson’s disease.

Our findings in CC, suggesting that tumors associated with high expression levels of ATP5A may be more amenable to respond to chemoradiation, await confirmation by others.

c-Jun N-terminal kinase stress-activated protein kinase (Jnk-Sapk1) (alias mitogen-activated protein kinase 8, Mapk8)

The Jnk-Sapk1 kinase acts as an integration point for multiple biochemical signals in a wide variety of cellular processes, such as pro-inflammatory cytokines or physical stress. Functional Jnk1 acts as a tumor suppressor and high levels of stromal expression have been associated with good outcomes and correlated with survival rates of NSCLC, CC, and HNSCC patients [86]. There is evidence suggesting that the JNK apoptotic signaling pathway can be inhibited by the PI3K-AKT pathway, suggesting that high expression of Jnk-Sapk1 protein may be indicative of low PI3K-AKT pathway activity.

• Glabridin, a flavonoid extracted from licorice (Glycyrrhiza glabra) has been shown to induce apoptosis of oral cancer cells through the JNK1/2 pathway and suggested as a potential therapeutic agent [70].

44-42 Mapk (alias p44 Erk1/Erk2)

Mapks transduce growth factors signals to mediate intracellular signaling. They are expressed at varying levels in all tissues and direct cellular responses to a wide range of stimuli, such as mitogens and pro-inflammatory cytokines. In a sample set of >100 CC, markers involved in the PI3K pathway (EGFR and phospho-Akt staining), correlated with high tumor grade and positive lymph node status (p<0.05), whereas significant expression of 44-42 MAPK was not a marker of poor outcome [87], coherent with the BioRAIDs series.

Dihydrofolate reductase (Dhfr)

Cancer cells use folate to meet their excess demand for nucleotides and amino acids. Methotrexate is an approved drug in recurrent CC in combination with Cisplatin or Tubulin interfering drugs. While chemoradiation appeared effective in the presence of high Dhfr protein levels [4], high Dhfr levels secondary to gene amplification have been correlated with methotrexate resistance in many cancers [71]. Amplifications of DHFR gene have been detected in premalignant cervical lesions [88].

• In preclinical assays, synergy has been shown by combining Methotrexate with Silmitasertib (CX-4945), a selective CK2 inhibitor.

Glycogen synthase kinase 3 (Gsk-3) α-β isoforms

Gsk-3 is a serine-threonine kinase, originally discovered as an important enzyme in glycogen metabolism, a negative regulator of glucose homeostasis, and-inhibitor of the Wnt/β-catenin pathway [72] by degradation mediated by β-catenin ubiquitin. The pro-survival BCL-2 family member MCL-1 can also be directly inhibited by Gsk-3 β [89]. Akt phosphorylates Gsk-3 isoforms at highly conserved N-terminal regulatory sites (Gsk-3 α-S21 and Gsk-3 β-S9), leading to kinase inactivation [90]. In HPV16 associated CC, a dramatic decrease in active Gsk-3 β-Try expression as well as ectopic overexpression of inactive Gsk-3 β-Ser have been associated with c-Myc expression [72]. In the BioRAIDs series, high levels of Gsk-3 α-β protein (activity status unknown) were associated with good outcomes, a finding awaiting further testing.

NADH ubiquinone oxidoreductase subunit A10 (NdufA10)

The protein encoded by this gene is a component of the 42 kDa complex I, the first enzyme complex in the mitochondrial electron transport chain. Ndufs are a large family with NADH dehydrogenase and oxidoreductase function and, together with the ATP5 genes family, are involved in mitochondrial energy production. Changes in NdufA10 methylation have been identified in human HNSCC tissue specimens from 2 geographically independent patient cohorts (Norway and UK).

Gata3 (GATA family of transcription factors; alias trans-acting T-cell-specific transcription factor)

Gata3 is a member of a family of transcription factors critical for the regulation of T-cell development and Th2 differentiation, involved in immune, allergic and inflammatory reactions, as well as in embryogenesis and cell differentiation and acts primarily as a tumor suppressor. Gata3 expression is considered a highly sensitive and specific marker for mesonephric/Wolffian remnants and for hyperplasia/tumors in the lower genital tract. In CC, thirty-eight percent of squamous cell carcinomas (SCCs) showed weak to moderate Gata3 staining in up to 50% of tumor cells [91].

P27 (alias cyclin-dependent kinase inhibitor 1b, Cdk1b)

P27 (CDKN1B gene product) represents a cyclin-dependent kinase inhibitor, which shares similarity with CDKN1A/p21. Under physiological conditions, the degradation of the P27 protein is required for the cellular transition from quiescence to proliferation. P27 binds and prevents activation of cyclin E-Cdk2 or cyclin D-Cdk4 complexes, thereby controlling cell cycle progression at G1.

• In cancer, P27 has been shown to be underexpressed in rapidly progressing oral SCC as well as in thyroid, colon, breast, prostate, and superficial bladder carcinomas. Genetic deletion of P27 has been shown in leukemias and pancreatic adenocarcinomas [92], in agreement with the preserved expression of P27 as a potential marker for good outcome.

Protein tyrosine phosphatase-non-receptor type 1 (Ptp1B)

Ptps are signaling molecules that catalyse the removal of phosphate groups from tyrosine residues, thereby acting as tumor suppressors. The role of Ptp1B in cancer physiopathology is controversial: it is frequently overexpressed in cancer due to genetic amplification of the chromosomal region 20q13, which has been associated with poor outcomes in multiple cancer types [93]. In BioRAIDs, high expression was associated with good control by chemoradiation consistent with its role as a tumor suppressor. While inhibition of Ptp1B appears counter-intuitive to its supposed role as tumor suppressor, several specific inhibitors are under development.

• Trodusquemine (MSI-1436) has been shown to inhibit tumor formation and abrogate metastasis in BC mouse models. Pterocarpans are isoflavonoids with inhibitory potency against many enzymes, including Ptp1B. A series of diethyl 2,5-diaminothiophene-3,4-dicarboxylate derivatives have been tested for their antiproliferative activity on cell lines and found to inhibit Ptp1B [94].

DISCUSSION

Many new cancer treatment strategies enter clinical trials with advanced stage, treatment-naïve CC patients receiving standard chemoradiation without pretreatment identification of targets of clinical significance (TOCS). BioRAIDs (NCT02428842) was a prospective biomarker trial, launched to address this issue and identified 25 TOCS (10 genetic, 15 protein), together with potential biomarkers for response to chemoradiation. Confirmation of these findings was ascertained in the cancer literature, except for protein post translationally modified protein forms which are currently not routinely assessed in clinical trials.

Emphasis in clinical trial design is on targeting dominant genetic alterations of known/suspected clinical significance. In CC, assumptions of potential targets based on cancer population frequency, suggested PIK3CA activation (40%) as the most prominent genetic TOCS. Interestingly, LoF gene alterations in CSMD3, UBR5, ARID1A, CBP, present in low frequency in a treatment naïve population (range: 5%–11%) ranked high by impact on outcome in a CoxBoost model, which included classical clinical parameters together with genetic alterations. When protein markers were introduced in this model, the following 3 protein forms outcompeted genetic markers for treatment resistance: 1 - a specific post translational modification in Pβ-cat-Ser675; 2 - high levels of the chaperone protein 14-3-3 β; and 3 - the inactive enzymatic form of Phospho-Eef2K-Ser366.

From the present perspective, new treatment options might address specifically these high-ranking protein alterations associated with poor outcome suggested by us and others, not excluding other treatment options acting on epigenetic, metabolic, immune, or protein degradation pathways. In the presence of resistance markers to platinum or radiation in refractory/relapsing disease, innovative targeted treatments may be based on antibodies, small molecules, or peptide mimetics that target PPI [24].

Targeting key proteins: While the literature abounds on reports linking β-catenin to outcome, the role of the precise molecular form Pβ-cat-Ser675 for poor outcome following chemoradiation had not been reported by others, to the best of our knowledge. Phosphorylation at the Ser-675 site has recently been shown to promote binding of β-catenin to CBP, leading to maintenance of stem/progenitor cell potency and self-renewal, while β-catenin binding to the coactivator p300, leads to activation of genes involved in differentiation. ICG-001 is a CBP/catenin antagonist, which specifically binds to CBP, but not to the histone acetylase p300 and was recently shown to promote dissociation between CBP and β-catenin [95]. ICG-001 is in clinical trials for sentinel lymph node mapping while ICG-001 derived clinical treatment forms (PRI-724 and OP-724-P101) are explored currently in clinical trials in solid, mostly digestive tract, cancers. Abundant protein 14-3-3 β or phospho-Eef2K-Ser366 are reported by us and others in association with outcome. Preclinical data for targeting these proteins have been reviewed [25,26], but presently no trials are listed on ClinicalTrials.gov website. Proteasome inhibitors (Bortezomib) have been suggested for tumors with high levels of the protein Phospho Eef2K-Ser366 or with genetic alterations in UBR5. Finally, HPV being the causal agent in >90% of CC, acting mainly through the inactivation of p53 and Rb by viral proteins E6 and E7, a potential reactivation of proteins p53 and Rb might be envisioned by combinations such as PRIMA-1 in association with Palbociclib.

Epigenetic targeting: KMT2D (MLL2) LoF epigenetic gene alterations, detected in 16% of primary CCs were found related to poor outcome following chemoradiation. The HDAC inhibitor Panobinostat was shown to restore tumor cell sensitivity to radiation and was suggested as a drug of interest for tumors with high levels of protein 14-3-3 β as well as to interrupt PPI between Pkm2 and β-cat. Other emerging epigenetic therapies suggest inhibition of the Bromodomain and Extra-Terminal domain (BET) protein BRD4 to hold great promise, together with DNA methyltransferases, EZH2, protein arginine N-methyltransferases and isocitrate dehydrogenases. BRD4 has been shown involved in the papillomavirus life cycle, as a co-factor for viral E2, mediating viral partitioning in some virus types highlighting the potential to target BET bromodomains to treat HPV infection [96].

Metabolic targeting of interest in CC may be directed against the following targets: Hsf1, Glut1, Pkm2 and Ca IX. Hsf1 when phosphorylated in Ser326 (Hsf1-pSer326) predicted response to bortezomib-containing chemotherapy in pediatric acute myeloid leukemia: a Children’s Oncology Group report [27]. Specific inhibitors are reviewed by Chatterjee and Burns [28]. Glut1, a glucose transporter could be inhibited by BAY-876 which specifically blocked growth of RB1-positive cancer in preclinical assays [47]. Pkm2 is both an enzyme in the cytoplasm where it regulates glucose metabolism and a nuclear transcription factor. NPD1008494 was shown to interrupt PPI between Pkm2 and β-catenin or Stat3, suppressing downstream signaling. A potential repurposable drug: Benserazide, a hexokinase-2 inhibitor (a constituent of Modopar), was recently withdrawn from market. In an ongoing clinical study, the role of Pkm2 detection in glioblastomas and healthy volunteers is further assessed (NCT03539731). Ca IX inhibitors are emerging agents for the treatment and imaging of hypoxic tumors [30]. Girentuximab, a chimeric radiolabeled antibody for the detection of Ca IX positive renal clear cell cancers is in clinical exploration.

Immune targeting combining anti Gitr and anti programmed cell death protein 1/Pd-l1 antibodies have been designed and combined reagents are in early clinical trials. This is of interest in CC where above median Gitr protein levels in the CC tumor extract were significantly associated with poor outcome, specifically since anti Pkm2 antibodies alone, in a non-stratified treatment-naïve population, did not improve PFS. Puzzlingly, high Ido expression in the CC tumor extract appeared positively associated with good outcome in some studies using chemotherapy [4], in disagreement with data suggesting Ido to be a poor prognostic marker through its role of immune cells tolerization in the tumor micro environment. Early reports on Ido inhibition did not to lead to clinical response at this stage. At least 12 clinical trials involving ICB are ongoing and a machine learning model to predict ICB response by integrating genomic, molecular, demographic and clinical data from a comprehensively curated cohort (MSK-IMPACT) with 1,479 patients treated with ICB across 16 may be assessed in future studies.

Tyrosine kinase pathway targeting: PIK3CA is altered/activated in close to 40% of advanced stage CC. Dual targeting of PIK3CA and epigenetic inhibition has been explored for tolerance (Fimepinostat).

CONCLUSION AND FUTURE PROSPECTS

Present treatment choices for individual CC patients in second/third line therapy rely too often on single candidate gene alterations, ignoring the presence of complex genetic alterations. They rarely take into account the countless protein dysfunctions.

Future focus might shift to the assessment and targeting of those markers which score highest fo their highest relevance on outcome in a larger population and adapt this information to the individual patient. Many phase 1/2 combination studies involving epigenetic and metabolic targeting are ongoing and appear of crucial interest to cater to the individual CC patient. The ClinicalTrials.gov-beta database lists more than 40 phase-1 and 76 “targeted” phase-2 studies under ‘malignant cervix uteri tumor’ while overall, CC patients still represent a minority subgroup in clinical study populations. CC remains a frequent disease, particularly in countries with minimal screening opportunities. It is of easy access for tumour sampling, which is vital for tumour board decision making and may serve as a model system for squamous cell cancers to be further assessed in a CC molecular platform trial.