Cancer Therapy

1. The Five Pillars of Modern Oncology

Cancer treatment in 2026 rests on five overlapping modalities. Few patients receive only one. Curative-intent regimens for solid tumours typically combine two or three; haematological malignancies frequently combine all five.

See Part I — Hallmarks for the biological rationale (each hallmark is a therapeutic target) and Part II — Genetic Basis for the driver-gene atlas that underwrites molecularly targeted therapy.

2. Cytotoxic Chemotherapy

The therapeutic principle: rapidly dividing cells are more vulnerable to DNA damage and mitotic disruption than quiescent cells. Cure depends on log-kill kinetics: each cycle kills a constant fraction of tumour cells, regardless of starting size.

\[ N(t) \;=\; N_0 \, e^{-k\,D\,t}, \qquad \text{cycles to cure} \;\sim\; \frac{\log(N_0)}{\log(\text{kill fraction})} \]

For a 1-kg tumour (~10¹² cells) and a 99% (2-log) kill per cycle, six cycles reduce the cell number to 10⁰ — the formal basis of multi-cycle MOPP, CHOP, ABVD, FOLFOX. The classes:

Alkylating agents

Transfer alkyl groups onto DNA bases (chiefly the N7 of guanine), producing monoadducts, intra- and inter-strand crosslinks. Replication forks collapse on encountering crosslinks; repair requires HR/FA pathways. Examples: cyclophosphamide (prodrug activated to phosphoramide mustard by hepatic CYP2B6; staple of CHOP, AC-T), ifosfamide, melphalan, busulfan, temozolomide (orally bioavailable, MGMT-substrate; standard for glioblastoma), dacarbazine (ABVD), bendamustine.

Platinum agents

Square-planar Pt(II) complexes that crosslink adjacent purines — principally the1,2-d(GpG) intrastrand adduct. Cisplatin (Rosenberg, 1965; FDA 1978) cures testicular cancer with bleomycin/etoposide (BEP); carboplatin is less nephro/ototoxic; oxaliplatin is the platinum of FOLFOX/FOLFIRINOX. Sensitivity tracks NER (ERCC1) and HR proficiency.

Antimetabolites

• 5-fluorouracil (5-FU) — thymidylate-synthase inhibitor (FdUMP-folate ternary complex). Backbone of CRC therapy (FOLFOX, FOLFIRI). DPD-deficient patients suffer life-threatening toxicity — pretreatment DPYD genotyping.
• Capecitabine — oral 5-FU prodrug.
• Methotrexate (MTX) — dihydrofolate-reductase inhibitor. ALL maintenance, CNS prophylaxis, osteosarcoma (high dose with leucovorin rescue).
• Pemetrexed — multi-target antifolate; standard in non-squamous NSCLC and mesothelioma.
• Gemcitabine — deoxycytidine analogue; pancreatic, NSCLC, bladder.
• Cytarabine (Ara-C) — cytidine analogue; AML “7+3” induction.
• 6-mercaptopurine, 6-thioguanine — purine analogues; ALL maintenance. TPMT genotyping mandatory.

Topoisomerase poisons

Stabilise the topoisomerase–DNA covalent intermediate, converting transient breaks into permanent ones. Etoposide and doxorubicin (and other anthracyclines: daunorubicin, idarubicin, epirubicin, mitoxantrone) target Topo II; irinotecan and topotecan target Topo I. Anthracyclines also intercalate and generate ROS — the source of their cardiotoxicity (cumulative dose limit, ∼450 mg/m² for doxorubicin; troponin / echo monitoring).

Microtubule poisons

• Taxanes (paclitaxel, docetaxel, nab-paclitaxel) — bind β-tubulin and stabilise microtubules, blocking spindle disassembly at metaphase. Breast, ovarian, NSCLC, prostate, head & neck.
• Vinca alkaloids (vincristine, vinblastine, vinorelbine) — bind tubulin and prevent polymerisation; vincristine is iconic in ALL and CHOP. Peripheral neuropathy is dose-limiting.
• Eribulin, ixabepilone — second-generation microtubule disruptors.

Other cytotoxics

Bleomycin (DNA strand breaks via radical chemistry; pulmonary fibrosis), L-asparaginase (depletes asparagine; ALL), hydroxyurea (RNR inhibitor; CML, MPN).

3. Targeted Small Molecules

Paul Ehrlich’s “magic bullet” (1908) — a chemical that finds and destroys disease without harming the host — was first realised in oncology by imatinib (Gleevec), the BCR-ABL inhibitor approved 2001 (Druker et al., NEJM 2001). Imatinib transformed CML from a 5-year-fatal disease into one with normal life expectancy on a daily oral pill, and became the canonical demonstration that a defined molecular driver can be drugged.

Kinase inhibitor classes

The protein kinome contains ∼520 kinases. ~80 small-molecule kinase inhibitors are FDA-approved (the majority for cancer). Mechanistic classes by binding mode:

• Type I — ATP-competitive, bind the active (DFG-in) conformation. Most reversible TKIs: erlotinib, gefitinib, dasatinib, crizotinib.
• Type II — ATP-competitive, bind the inactive (DFG-out) conformation, exploiting an allosteric pocket. Imatinib, sorafenib, nilotinib.
• Type III (allosteric) — bind outside the ATP pocket. Trametinib (MEK), asciminib (ABL myristoyl pocket).
• Type IV (covalent) — form covalent bonds, usually with a non-catalytic cysteine. Afatinib, osimertinib (Cys797 of EGFR), ibrutinib (Cys481 of BTK), sotorasib (Cys12 of KRAS-G12C).
• Pseudo-irreversible — very long residence time without covalency.

Landmark targeted drugs

See Part IV — DNA Repair for the synthetic-lethal logic of PARP inhibitors in BRCA1/2-deficient cancers, and the Pharmacology course for kinase-inhibitor PK/PD principles.

4. Antibody Therapy

Köhler & Milstein’s 1975 hybridoma technology made monoclonal antibodies possible; humanisation (chimeric, then fully human via phage display or transgenic mice) made them clinically tolerable. Therapeutic antibodies in oncology fall into three architectural classes.

A. Naked monoclonal antibodies

Bind a tumour antigen and kill via Fab-mediated signalling blockade and Fc-mediated effector functions: ADCC (antibody-dependent cellular cytotoxicity, via FcγRIIIa on NK cells), ADCP (phagocytosis by macrophages), CDC (complement-dependent cytotoxicity).

• Trastuzumab (Herceptin, anti-HER2/ERBB2) — HER2-amplified breast and gastric cancer. Approved 1998 on the basis of Slamon’s pivotal NEJM 2001 trial; transformed HER2+ disease from worst to best prognosis subtype. Adjuvant trastuzumab (HERA, BCIRG-006) cuts recurrence by ~50%.
• Pertuzumab — anti-HER2 dimerisation domain II; complementary to trastuzumab. CLEOPATRA: trastuzumab + pertuzumab + docetaxel in metastatic HER2+ breast.
• Rituximab — anti-CD20; B-cell NHL, CLL. R-CHOP is the standard for DLBCL. The first FDA-approved monoclonal for cancer (1997).
• Cetuximab / panitumumab — anti-EGFR; metastatic CRC (KRAS/NRAS-wild-type only), head & neck SCC.
• Bevacizumab — anti-VEGF-A; deprives tumour vasculature of pro-angiogenic signal. CRC, NSCLC, GBM, ovarian. Folkman’s anti-angiogenic hypothesis vindicated.
• Daratumumab — anti-CD38; multiple myeloma.
• Obinutuzumab — glycoengineered anti-CD20 with enhanced ADCC; CLL, follicular lymphoma.

B. Antibody-drug conjugates (ADCs)

An antibody is the homing missile, a cytotoxic warhead the payload, and a chemical linker connects them — designed to be stable in plasma but cleaved inside the target cell. ADCs deliver chemotherapy intracellularly only to antigen-expressing cells, dramatically improving the therapeutic index.

• Trastuzumab deruxtecan (T-DXd, Enhertu) — anti-HER2 IgG1 + topoisomerase-I inhibitor (deruxtecan, DXd) via cleavable tetrapeptide linker; DAR ≈ 8. Active in HER2-low breast cancer (DESTINY-Breast04, NEJM 2022) — redefined the population of patients who benefit from HER2 targeting.
• Brentuximab vedotin (Adcetris) — anti-CD30 + MMAE (auristatin, microtubule poison); Hodgkin lymphoma, ALCL.
• Sacituzumab govitecan (Trodelvy) — anti-Trop-2 + SN-38 (active metabolite of irinotecan); triple-negative breast cancer (ASCENT).
• Enfortumab vedotin — anti-Nectin-4 + MMAE; urothelial cancer.
• Polatuzumab vedotin — anti-CD79b + MMAE; DLBCL.
• Gemtuzumab ozogamicin — anti-CD33 + calicheamicin; AML. The first ADC, withdrawn 2010, re-approved 2017 at a different dose.

C. Bispecific antibodies and BiTEs

Engineered to bind two different epitopes simultaneously. The dominant oncology design is the T-cell engager: one arm binds CD3 on a T-cell, the other binds a tumour antigen, forcing immunological synapse formation regardless of TCR specificity.

• Blinatumomab — CD3 × CD19 BiTE; B-ALL. The first approved T-cell engager (2014). Continuous IV infusion; short half-life; CRS toxicity.
• Teclistamab, elranatamab — CD3 × BCMA bispecifics; multiple myeloma.
• Talquetamab — CD3 × GPRC5D; multiple myeloma.
• Tarlatamab — CD3 × DLL3 BiTE; small-cell lung cancer (DeLLphi-301).
• Mosunetuzumab, glofitamab, epcoritamab — CD3 × CD20 bispecifics; B-cell NHL.
• Amivantamab — EGFR × MET bispecific; EGFR exon-20 NSCLC.

5. Immune Checkpoint Blockade

Among modern cancer therapies this is the revolution. James Allison and Tasuku Honjo shared the 2018 Nobel Prize in Physiology or Medicine for the discovery that T-cell activity is restrained by inhibitory receptors — CTLA-4 and PD-1 — and that blockade of these receptors with antibodies releases anti-tumour immunity. The clinical consequence: durable, sometimes curativeresponses in metastatic cancers (melanoma, NSCLC, MSI-H tumours) where five-year survival was previously near zero.

The molecular logic

T-cell activation requires two signals: TCR engagement of peptide-MHC (signal 1) and co-stimulation through CD28 binding B7 (signal 2). This must be balanced. Two inhibitory checkpoints turn off T-cell responses to prevent autoimmunity:

• CTLA-4 — upregulated on activated T-cells, binds B7 with higher affinity than CD28 and outcompetes co-stimulation. Acts in lymph nodes during T-cell priming.
• PD-1 — expressed on activated and exhausted T-cells. On engagement of its ligands PD-L1 (B7-H1) or PD-L2 on tumour cells, antigen-presenting cells, or stromal cells, it recruits SHP-2 phosphatase to the TCR signalosome, dephosphorylating CD3ζ and CD28 and shutting down T-cell function. Acts in peripheral tissues during effector phase.

Tumours co-opt PD-L1 expression as a way to disarm tumour-infiltrating lymphocytes. Blocking the PD-1·PD-L1 interaction with antibody releases T-cell killing of tumours that the immune system already “sees.”

Approved checkpoint inhibitors

Landmark trials

• Hodi NEJM 2010 — ipilimumab in metastatic melanoma; first phase III to show survival benefit, ever, in this disease.
• KEYNOTE-024 (NSCLC) — pembrolizumab vs platinum doublet, PD-L1 TPS ≥50%; median OS 30 vs 14 months. Established checkpoint monotherapy as first-line.
• CheckMate-067 (melanoma) — ipilimumab + nivolumab vs each alone; 5-year OS 52% with the combination — transformative in a previously fatal disease. Toxicity (grade-3+ irAEs ~60%) is the price.
• KEYNOTE-189 — pembrolizumab + chemo in non-squamous NSCLC (chemo-IO combinations).
• PACIFIC — durvalumab consolidation in unresectable stage III NSCLC after chemoradiation.
• KEYNOTE-A18 / NICHE — neoadjuvant immunotherapy in MMR-deficient colon cancer; pathological CR rates near 100%.

Response biomarkers

• PD-L1 IHC — tumour proportion score (TPS) or combined positive score (CPS); imperfect predictor (heterogeneous, dynamic).
• Tumour mutational burden (TMB) — high TMB → more neoantigens → better response. TMB-H (≥10 mut/Mb) earned a tissue-agnostic pembrolizumab approval (2020).
• MSI-H / MMR-deficient — first tissue-agnostic indication (2017). Lynch-syndrome and sporadic-MMR-d cancers respond exquisitely.
• Inflamed gene signatures — T-cell-inflamed phenotype, IFN-γ signatures.
• Gut microbiome composition — faecal-microbiota transplantation rescues anti-PD-1 non-responders in some studies.

Immune-related adverse events (irAEs)

The other side of releasing T-cell activity is autoimmunity. Common irAEs: dermatitis (rash, vitiligo), colitis, hepatitis, pneumonitis, endocrinopathies (thyroiditis, hypophysitis, type-1 diabetes), nephritis, myocarditis (rare but fulminant; combination >monotherapy). Management: hold drug, glucocorticoids; for refractory or severe cases, infliximab, vedolizumab, mycophenolate. Endocrine irAEs are usually permanent (lifelong replacement). Surveillance is part of every checkpoint regimen.

6. Adoptive Cell Therapy

Rather than coaxing the patient’s own T-cells into action, adoptive cell therapy manufactures the cancer-fighting cell ex vivo and infuses it back. The flagship modality — CAR-T cells— produces unprecedented cure rates in B-cell malignancies and is the first cellular gene therapy in oncology to be approved (tisagenlecleucel, 2017).

CAR construct anatomy

A chimeric antigen receptor (CAR) is a synthetic single-protein receptor that fuses an antibody-derived antigen-binding domain to T-cell signalling machinery. The canonical second-generation CAR has five regions:

• scFv — single-chain variable fragment derived from a tumour-antigen-specific monoclonal antibody (e.g. anti-CD19 from mAb FMC63). Provides MHC-independent antigen recognition.
• Hinge / spacer — usually IgG4 or CD8α hinge; permits antigen reach.
• Transmembrane domain — CD8α or CD28; anchors the CAR.
• Costimulatory domain — 4-1BB (CD137) or CD28 cytoplasmic tail. 4-1BB CARs (tisagenlecleucel) persist longer; CD28 CARs (axicabtagene ciloleucel) act faster but exhaust sooner.
• CD3ζ activation domain — provides signal 1 (TCR-equivalent) on antigen engagement.

\[ \text{scFv} - \text{hinge} - \text{CD8 TM} - \text{4\text{-}1BB} - \text{CD3}\zeta \quad (\text{tisagenlecleucel}) \]

First-generation CARs (CD3ζ alone, 1990s) failed for lack of persistence. Second generation (with one costimulatory domain) drove the modern revolution. Third generation (two costimulatory domains: 4-1BB + CD28) and armoured CARs (cytokine co-expression, e.g. IL-12, IL-18) are in development.

Manufacturing workflow

1. Leukapheresis — harvest patient T-cells (autologous) from peripheral blood.
2. T-cell selection & activation — CD4/CD8 enrichment, anti-CD3/CD28 bead stimulation.
3. CAR transduction — lentiviral or gammaretroviral vector encoding the CAR construct.
4. Expansion — 9–14 days in IL-2 / IL-7 / IL-15.
5. Lymphodepleting chemotherapy — fludarabine + cyclophosphamide given to patient to make immunological space and deplete regulatory T-cells.
6. Infusion — CAR-T cells expand in vivo to 10⁵-fold and traffic to malignant B-cell compartments.

Approved CAR-T products

Landmark CAR-T trials

• Maude / Grupp NEJM 2014 (CHOP-007) — CTL019 (tisagenlecleucel) in relapsed paediatric ALL; Carl June’s landmark cure of Emily Whitehead. Complete remission rate >80% in patients with otherwise inevitable death.
• ZUMA-1 (axi-cel in DLBCL) — ORR 82%, CR 54%, median OS 25.8 months in chemo-refractory disease.
• ZUMA-7 / TRANSFORM — CAR-T as second-line in DLBCL, beating standard-of-care salvage chemo + autologous stem cell transplant.
• CARTITUDE-1 — cilta-cel in heavily pre-treated multiple myeloma; CR 83%.

Toxicities unique to cellular therapy

• Cytokine release syndrome (CRS) — massive release of IL-6, IFN-γ, TNF-α from activated T-cells and stimulated macrophages. Fever, hypotension, hypoxia, capillary leak. Graded 1–4. Treated with tocilizumab (anti-IL-6R) ± corticosteroids. Onset typically days 2–7 post-infusion.
• Immune effector cell-associated neurotoxicity syndrome (ICANS) — tremor, dysphasia, encephalopathy, seizures, occasionally fatal cerebral oedema. Mechanism: blood-brain-barrier breakdown, endothelial dysfunction, cytokine penetration. Treated with corticosteroids; tocilizumab does not penetrate CNS well.
• B-cell aplasia — on-target, off-tumour effect of CD19 CAR-T; lifelong IVIG replacement may be needed.
• Prolonged cytopenias / haemophagocytic lymphohistiocytosis (HLH-like).
• Secondary T-cell malignancies — rare, related to insertional mutagenesis from lentiviral vector. FDA boxed warning 2024.

Other adoptive modalities

• Tumour-infiltrating lymphocytes (TIL) — T-cells extracted from a resected tumour, expanded ex vivo with high-dose IL-2, infused back. Lifileucel (Amtagvi, 2024): first FDA-approved TIL therapy — advanced melanoma post-checkpoint failure. The Steven Rosenberg legacy.
• TCR-T (T-cell receptor engineered T-cells) — recognises intracellular antigens via MHC; can target tumour-restricted self-peptides. Afami-cel (afamitresgene autoleucel, 2024): first TCR-T approved (synovial sarcoma, MAGE-A4). Promise: solid tumours.
• Allogeneic / off-the-shelf CAR-T — donor-derived, gene-edited (e.g. TRAC knockout) to prevent GvHD; under development.
• CAR-NK cells — lower CRS, allogeneic possible; in trials.

7. Radiotherapy

Ionising radiation kills cells by inducing DNA damage — chiefly double-strand breaks, partly through direct ionisation of DNA, partly through radiolytically generated hydroxyl radicals (the “indirect effect,” oxygen-dependent). The lethal lesion is the unrepaired or misrepaired DSB. See Part IV — DNA Repair for the molecular machinery whose failure makes radiation curative.

\[ S(D) \;=\; e^{-(\alpha D + \beta D^2)} \quad \text{(linear-quadratic survival model)} \]

The linear-quadratic model is the workhorse of clinical radiobiology: α reflects single-track lethal damage, β reflects accumulation of sublethal damage. Tumours and late-responding normal tissue have different α/β ratios — fractionation schedules exploit this differential to deliver tumoricidal dose with tolerable toxicity. The “5 R’s” of radiobiology: repair, redistribution, repopulation, reoxygenation, radiosensitivity.

Modalities

• Conventional 3D-CRT — CT-planned external-beam photon radiotherapy.
• IMRT (intensity-modulated) — sculpted dose distributions through dynamic multi-leaf collimators; spares OARs (organs at risk).
• VMAT — volumetric arc therapy; faster delivery than IMRT.
• SBRT / SRS — stereotactic body / cranial radiosurgery; ablative single-fraction or few-fraction regimens for small targets (early NSCLC, brain metastases, oligometastases).
• Proton therapy — charged-particle dose deposition with the Bragg peak; reduces integral dose and entrance/exit dose to normal tissue. Particularly useful in paediatrics and skull-base/spinal targets.
• Carbon-ion therapy — higher LET, higher RBE; under investigation for radioresistant tumours.
• FLASH radiotherapy — ultra-high dose rates (>40 Gy/s) that paradoxically spare normal tissue while preserving tumour kill; mechanism debated (oxygen depletion, radical recombination). Early clinical translation.
• Brachytherapy — sealed radioactive sources placed in or adjacent to tumour (cervix, prostate, breast).
• Theranostics — targeted radioligand therapy; lutetium-177 dotatate (Lutathera, somatostatin-receptor neuroendocrine), lutetium-177 PSMA-617 (Pluvicto, prostate cancer; VISION trial).

Curative-intent radiotherapy is standard for early-stage prostate, head & neck, cervical, anal, and Hodgkin lymphoma. Combined chemoradiation cures locally advanced cervical, anal, oesophageal, and stage-III NSCLC. Palliative radiotherapy controls pain (bone mets), bleeding, obstruction, and brain metastases.

8. Hormone Therapy

Cancer endocrinology is the original targeted therapy. In 1896, George Beatson reported that oophorectomy could induce regression of metastatic breast cancer — six decades before the oestrogen receptor was discovered. Charles Huggins won the 1966 Nobel for showing that orchiectomy regresses prostate cancer (1941). Both observations are now executed pharmacologically.

Breast cancer (HR+ disease, ~70%)

• Selective oestrogen-receptor modulators (SERMs) — tamoxifen: ER antagonist in breast, agonist in bone/endometrium. 5–10 years adjuvant tamoxifen in pre-menopausal HR+ disease.
• Aromatase inhibitors (AIs) — anastrozole, letrozole, exemestane. Block conversion of androgens to oestrogens in peripheral tissues; effective only in post-menopausal women.
• Selective oestrogen-receptor degraders (SERDs) — fulvestrant (IM injection), elacestrant (oral, ESR1-mutant disease).
• GnRH agonists — goserelin, leuprolide; medical ovarian suppression.
• CDK4/6 inhibitors (palbociclib, ribociclib, abemaciclib) added to endocrine therapy — doubled PFS in HR+ HER2− metastatic disease (PALOMA-2, MONALEESA-2).

Prostate cancer (androgen-dependent)

• GnRH agonists / antagonists — leuprolide, degarelix, relugolix; medical castration (testosterone <50 ng/dL).
• Androgen-receptor antagonists — enzalutamide, apalutamide, darolutamide; second-generation, more potent, brain-penetrant (with seizure risk).
• CYP17 inhibitor — abiraterone (with prednisone); blocks adrenal/intratumoural androgen synthesis. COU-AA-302, LATITUDE.
• PARP inhibitors in BRCA1/2-mutated metastatic CRPC (PROfound, olaparib).
• ¹⁷⁷Lu-PSMA-617 radioligand — VISION trial, post-novel-hormonal therapy.

Endometrial cancer

Progestins (medroxyprogesterone, megestrol) for low-grade endometrioid disease.

9. Drug Resistance

The central problem of clinical oncology, after the initial response, is resistance. Cancer is an evolutionary system: drug exposure is a selective pressure, and tumours with their high mutation rates and large clonal populations almost always harbour pre-existing or evolvable resistant subclones.

Primary vs acquired resistance

Primary (intrinsic) resistance: tumour never responds. Often reflects absence of the predicted target dependence (e.g. EGFR-mutant NSCLC with concurrent KRAS) or lack of biomarker. Acquired resistance: tumour shrinks then grows back. Always evolutionary — either pre-existing minor clone selection or de novo mutation under drug pressure.

Mechanisms

• Gatekeeper mutations — bulky residue at the bottom of the ATP pocket sterically excludes drug. EGFR T790M after erlotinib/gefitinib (~50% of acquired resistance) — addressed by osimertinib (covalent at C797S). ABL T315I after imatinib — addressed by ponatinib, asciminib (allosteric).
• On-target amplification — gene amplification of the target (BCR-ABL in CML; MET in EGFR-NSCLC).
• Bypass-pathway activation — alternative receptor activation maintains downstream signalling (MET amplification, HER2 amplification, ERBB3 upregulation in EGFR-mutant NSCLC; NRAS-mutation rescue of BRAF-inhibited melanoma).
• Lineage plasticity / phenotypic switching — small-cell transformation in EGFR-NSCLC; neuroendocrine differentiation in CRPC.
• Drug efflux / metabolism — P-glycoprotein, MRP1; CYP-mediated inactivation.
• Tumour-microenvironment-mediated resistance — stromal HGF, IL-6, exosomal transfer; CAFs.
• Persister cells — non-genetic, transient drug-tolerant state; epigenetic, often reversible. Recently recognised as the seed of acquired genetic resistance (Sharma et al., Cell 2010).
• Immune escape — B2M / HLA loss, PD-L1 / IFN-pathway loss; resistance to checkpoint blockade.

Clonal evolution

Liquid-biopsy ctDNA monitoring reveals the dynamics in real time: a single dominant T790M clone emerging months before radiographic progression on first-line erlotinib; polyclonal MEK1/2 mutations after MEK inhibition; heterogeneous CDK4/6 resistance including RB1 loss, FAT1, AURKA. Treatment now targets evolution itself: switch to osimertinib at the first ctDNA T790M signal.

10. Combination Strategies

Cancer cures require combinations: this has been true since MOPP cured Hodgkin lymphoma in 1965. The principle (DeVita) was to combine drugs with non-overlapping toxicities and non-cross-resistant mechanisms, each at full single-agent dose. The same logic underwrites every modern curative regimen.

Patterns of combination

• Concurrent — chemo + RT in head & neck, anal, cervical, NSCLC stage III; chemo + IO in NSCLC (KEYNOTE-189); BRAF + MEK in melanoma (COMBI-d).
• Sequential — neoadjuvant chemo → surgery → adjuvant chemo ± RT ± targeted therapy in HER2+ breast.
• Maintenance — ongoing low-toxicity therapy after induction; PARPi maintenance in ovarian (SOLO1); pembrolizumab maintenance in NSCLC.
• Switch maintenance — introduce new agent after induction with another.
• Adjuvant / consolidation — durvalumab after chemoRT (PACIFIC); osimertinib adjuvant in EGFR+ resected NSCLC (ADAURA).

Rational design

• Vertical pathway blockade — BRAF + MEK; HER2 + tucatinib; EGFR + downstream MEK.
• Horizontal pathway blockade — PI3K + MEK (limited by toxicity).
• Synthetic lethality — PARPi in HRD; DNA-damaging chemo + ATM/ATR inhibition; KRAS-G12C + SHP2.
• Chemo-immunotherapy — chemotherapy releases tumour antigens and modulates the microenvironment; checkpoint blockade then drives durable response.
• Doublet checkpoint blockade — ipilimumab + nivolumab; nivolumab + relatlimab. Greater activity, greater toxicity.

Combinations also raise the toxicity bar: regulatory and clinical decision-making is fundamentally about which combinations have non-overlapping toxicity profiles. Hepatotoxicity, pneumonitis, cardiac toxicity and myelosuppression can all be compounded by ill-considered combinations.

11. Personalised / Precision Oncology

The genomic atlas (Part II) created the substrate; the targeted-therapy era exploits it. Modern oncology routinely sequences metastatic tumours with broad next-generation sequencing panels (FoundationOne, MSK-IMPACT, Tempus) at first diagnosis, querying hundreds of genes and gene-fusions in a single test.

Companion diagnostics

Many targeted drugs have FDA-mandated companion diagnostic tests: HER2 IHC/FISH for trastuzumab; EGFR sequencing for erlotinib; BRAF V600E for vemurafenib; PD-L1 IHC for pembrolizumab in NSCLC; BRCA1/2 sequencing for olaparib; MSI testing for MMR-deficient tissue-agnostic indications.

Tissue-agnostic indications

• Pembrolizumab — MSI-H/dMMR (2017), TMB-H ≥10 mut/Mb (2020).
• Larotrectinib, entrectinib — NTRK1/2/3 fusions (any solid tumour).
• Selpercatinib — RET fusions.
• Dostarlimab — dMMR (2021); spectacular pCR results in MMR-deficient rectal cancer (Cercek et al., NEJM 2022 — 100% complete responses, no surgery needed).

Modern trial designs

• Basket trials — one drug across many tumour types sharing a molecular alteration (NCI-MATCH, larotrectinib NTRK).
• Umbrella trials — one tumour type, multiple molecularly defined arms (Lung-MAP, I-SPY).
• Platform / adaptive trials — arms added and dropped based on interim Bayesian analyses; allow rapid drug development across heterogeneous biology.
• Real-world evidence — large electronic-health-record cohorts (Flatiron, GENIE) increasingly support regulatory decisions.

Liquid biopsy

Circulating tumour DNA (ctDNA) sequencing from a tube of blood replaces or complements tissue genotyping when biopsy is impossible. Approved uses: EGFR genotyping in NSCLC, BRCA reversion-mutation tracking in ovarian, minimal-residual-disease detection (Signatera, Reveal) post-curative-intent surgery in CRC and breast.

12. Future Directions

Cancer therapy in 2026 is in a golden age, but the open frontiers are clear and multiple. The next decade will likely see:

• Bispecific T-cell engagers across solid tumours — tarlatanab (DLL3 in SCLC), HLE-BiTEs with longer half-lives, masked / pro-drug bispecifics activated in tumour microenvironment.
• KRAS pan-inhibitors — G12D (RMC-9805), G12V, multi-RAS “dark KRAS” agents; KRAS-multi-on covalent inhibitors. Pancreatic cancer — the great unmet need — is the prize.
• Targeted protein degradation — PROTACs and molecular glues drug previously undruggable targets (transcription factors, scaffold proteins) by recruiting an E3 ligase to the target. ARV-471 (vepdegestrant, ER) and KT-474 (IRAK4) lead the clinical field.
• mRNA cancer vaccines — personalised neoantigen vaccines from tumour exome sequencing. mRNA-4157 (BioNTech / Moderna) plus pembrolizumab in resected high-risk melanoma — KEYNOTE-942 phase II showed dramatic recurrence reduction.
• TCR-T for solid tumours — afami-cel (synovial sarcoma) is just the beginning; NY-ESO-1, MAGE-A4, HPV-E6/E7 TCRs are advancing in cervical, head & neck, and other tumours.
• In-vivo CAR generation — lipid-nanoparticle delivery of CAR mRNA to T-cells in vivo (no leukapheresis, no manufacturing). Capstan, Umoja, Orna.
• AI-guided drug design — AlphaFold-derived structures, generative chemistry, ML-driven biomarker discovery; protein-language-model-based antibody design.
• Multi-cancer early-detection (MCED) — ctDNA methylation and fragmentomics screen for multiple cancers from a single blood draw (Galleri, Pathfinder); the screening modality of the future.
• Synthetic-lethal combinations beyond PARP — ATR, WEE1, USP1, POLθ inhibitors in HRD; PRMT5 in MTAP-deleted cancers.
• Microbiome modulation — FMT and engineered probiotic adjuncts to checkpoint blockade.

The story since Druker’s NEJM 2001 imatinib paper is one of accelerating molecular precision with simultaneously expanding biological reach: from single-driver kinase inhibitors, through monoclonal antibodies and ADCs, to checkpoint blockade and cellular therapy. Each modality addresses a distinct hallmark (Part I), and the most successful regimens combine modalities. The remaining hard problems — metastasis, the cold tumour, persister cells, RAS in pancreatic cancer, MYC — will yield to the same logic, more refined.