Key papers

In AMPLIFY, an all-oral, 14-month course of acalabrutinib plus venetoclax beat FCR or BR chemo-immunotherapy in fit CLL patients, offering a time-limited alternative to indefinite pills.

Adding a BRAF inhibitor and an EGFR antibody to first-line chemotherapy roughly doubled survival in BRAF V600E-mutated metastatic bowel cancer, one of the worst-prognosis subtypes.

In patients who were transplant-ineligible or deferred transplant, the daratumumab quadruplet raised deep-remission rates from about 39% to 61% and cut progression risk by 43%.

In CHALLENGE, three years of supervised, goal-based exercise after adjuvant chemotherapy for colon cancer reduced recurrence or death by 28% and deaths by 37%, the first randomised proof that exercise itself improves cancer survival.

Ten-year follow-up of the trial that established combination checkpoint blockade shows median survival over six years with nivolumab plus ipilimumab and a plateau in the survival curve consistent with cure.

Ten years after starting treatment, about half of patients with advanced melanoma treated with nivolumab plus ipilimumab were still alive, most without any further treatment, showing that immunotherapy can cure a disease that once killed most patients within a year.

Two courses of blinatumomab added to standard chemotherapy cut relapses in children with average- or higher-risk standard-risk leukaemia, raising three-year disease-free survival from 88% to 96%.

Patients whose blood turned positive for tumour DNA after cystectomy lived longer with atezolizumab than placebo; patients who stayed ctDNA-negative did well without any treatment. It is the first ctDNA-guided adjuvant trial to improve survival.

A small pretzel-shaped device that slowly releases gemcitabine inside the bladder cleared carcinoma in situ in about four out of five patients whose cancer had stopped responding to BCG, offering an alternative to bladder removal.

For small-cell lung cancer confined to the chest, adding two years of durvalumab after chemoradiotherapy extended median survival from under three years to over four and a half, the first advance in this setting in decades.

Given as the first chemotherapy-type treatment after hormone therapy stopped working, trastuzumab deruxtecan delayed progression by about five months compared with standard chemotherapy, including in tumours with barely detectable HER2.

Giving the bispecific blinatumomab to adults whose leukaemia was already undetectable after chemotherapy raised three-year survival from 68% to 85%.

Combining the Nectin-4 antibody-drug conjugate enfortumab vedotin with pembrolizumab nearly doubled survival compared with platinum chemotherapy in advanced urothelial cancer, the biggest advance in this disease in 40 years.

In IMerge, imetelstat, the first telomerase inhibitor to reach approval, freed about 40% of heavily transfused MDS patients from transfusions for eight weeks and 28% for six months, versus 15% and 3% with placebo.

Adding pembrolizumab to standard chemoradiotherapy for high-risk locally advanced cervical cancer reduced progression by 30% and later improved survival, the first systemic advance in this setting since cisplatin was added to radiotherapy in 1999.

Combining an EGFR-MET bispecific antibody with a third-generation EGFR pill beat osimertinib alone, delaying progression by about seven months and later improving survival, at the cost of more side effects.

Giving just two cycles of combination immunotherapy before removing melanoma lymph nodes cut relapses by more than two-thirds compared with a year of standard immunotherapy after surgery, and most patients needed no further treatment.

Three years of the CDK4/6 inhibitor ribociclib added to hormone therapy reduced relapses in stage II-III hormone-receptor-positive breast cancer, including some node-negative patients.

Using the radioactive drug lutetium dotatate as the first treatment for faster-growing neuroendocrine tumours, rather than saving it for later, nearly tripled the time without progression compared with high-dose octreotide.

Adding the immunotherapy durvalumab to chemotherapy before bladder removal, and continuing it afterwards, reduced relapse and death in muscle-invasive bladder cancer, the first improvement on neoadjuvant chemotherapy in two decades.

Two doses of immunotherapy over four weeks before surgery left little or no living tumour in 95% of patients with mismatch-repair-deficient colon cancer, and none had relapsed at three years.

In NICHE-2, four weeks of checkpoint blockade before surgery eliminated nearly all tumour in 95% of patients with mismatch-repair-deficient colon cancer, and none had relapsed at three years.

Adding daratumumab to the standard three-drug induction, transplant and maintenance cut progression or death by 58% and pushed MRD-negativity to three-quarters of patients.

T cells engineered with a receptor recognising the MAGE-A4 cancer antigen shrank tumours in 37% of patients with advanced synovial sarcoma or myxoid liposarcoma, leading to the first approval of a TCR-T therapy.

T cells taken from patients and engineered to recognise the MAGE-A4 protein shrank tumours in about four in ten patients with advanced synovial sarcoma, leading to the first approval of a TCR T-cell therapy for any solid cancer.

Adding a PD-1 antibody to chemotherapy beat the brentuximab-based standard, with 92% of patients progression-free at two years and less nerve damage.

Blocking menin, a scaffold protein the leukaemia depends on, produced remissions in heavily pretreated patients with KMT2A-rearranged or NPM1-mutated acute leukaemia.

Moving BCMA CAR-T to the second line cut the risk of progression or death by 74% compared with standard triplets, and later improved overall survival.

The first drug to directly block mutant KRAS beat docetaxel chemotherapy on delaying progression in KRAS G12C lung cancer, but only by about a month, and did not improve survival.

Combining a KRAS G12C inhibitor with an EGFR antibody produced responses in about a quarter of patients with heavily pretreated KRAS G12C bowel cancer, versus none with standard chemotherapy, and more than doubled the time to progression.

In COMMANDS, luspatercept, a drug that frees late-stage red cell production from TGF-beta-family braking, freed 59% of transfusion-dependent MDS patients from transfusions for at least 12 weeks, against 31% with the standard erythropoietin injection.

A bispecific antibody that pulls T cells onto small-cell lung cancer cells produced responses in 40% of patients whose cancer had come back after chemotherapy, lasting far longer than any previous drug in this setting.

In EPCORE NHL-1, the off-the-shelf bispecific antibody epcoritamab produced responses in 63% of patients with relapsed aggressive lymphoma, including many whose CAR-T had failed.

An oral drug that blocks the mutant IDH enzyme more than doubled the time before slow-growing IDH-mutant brain tumours progressed, letting patients postpone radiotherapy and chemotherapy for years.

The first randomised CAR-T trial in myeloma tripled median progression-free survival (13.3 versus 4.4 months) compared with standard combinations after two to four prior lines.

Elranatamab produced responses in 61% of heavily pretreated myeloma patients and showed that dosing can be thinned to every two weeks once patients respond.

A JAK inhibitor that also blocks the anaemia-driving ACVR1 pathway improved symptoms and spleen size without worsening, and often improving, anaemia in previously treated patients.

A four-drug chemotherapy combination built on liposomal irinotecan extended survival in metastatic pancreatic cancer by about two months compared with the most widely used doublet, the first positive first-line trial in a decade.

Adding the FLT3 inhibitor quizartinib to standard chemotherapy, and continuing it for up to three years, roughly doubled median survival in FLT3-ITD acute myeloid leukaemia.

Adding the PD-1 antibody dostarlimab to first-line chemotherapy for advanced endometrial cancer cut progression by 72% in tumours with defective mismatch repair and by about a third overall, and later improved survival.

In stomach cancers that express the protein Claudin 18.2, adding the antibody zolbetuximab to chemotherapy extended survival by nearly three months, making Claudin 18.2 a new biomarker to test for.

Adding the IDH1 inhibitor ivosidenib to azacitidine tripled median survival, from 7.9 to 24 months, in older patients with IDH1-mutated AML.

All 12 patients with locally advanced dMMR rectal cancer had a complete clinical response to a PD-1 antibody alone, avoiding chemotherapy, radiotherapy and surgery.

Just three cycles of chemotherapy with the immunotherapy nivolumab before surgery wiped out all viable tumour in a quarter of patients and reduced relapse or death by about a third, without making surgery harder.

A newer antibody-drug conjugate, trastuzumab deruxtecan, kept HER2-positive metastatic breast cancer under control roughly four times longer than the previous standard, T-DM1, and later lengthened survival.

Breast cancers with only a little HER2 on their surface, long called HER2-negative, responded to trastuzumab deruxtecan and patients lived about six months longer than on chemotherapy. It created the HER2-low category.

Six months of the PD-1 antibody dostarlimab made every tumour disappear in a small group of patients with mismatch-repair-deficient rectal cancer, allowing them to avoid chemotherapy, radiotherapy and surgery.

Giving chemotherapy only to patients with tumour DNA in their blood after surgery cut chemotherapy use from 28% to 15% with no loss in recurrence-free survival at two years.

Adding the immunotherapy pembrolizumab to chemotherapy before surgery, then continuing it afterwards, cut the risk of relapse or death by about a third in stage II-III triple-negative breast cancer and later improved survival.

A ready-made antibody that pulls T cells onto myeloma cells produced responses in 63% of patients after a median of five prior therapies, without the wait for cell manufacturing.

Replacing one chemotherapy drug in R-CHOP with a CD79b antibody-drug conjugate modestly reduced progression (2-year PFS 76.7% versus 70.2%) without changing survival or side effects.

Adding an antibody against a second immune brake, LAG-3, to nivolumab delayed progression in advanced melanoma compared with nivolumab alone, with far fewer serious side effects than the ipilimumab combination.

In a head-to-head trial, tumour-infiltrating lymphocyte therapy halved the risk of progression compared with ipilimumab in melanoma that had mostly already failed PD-1 blockade.

In TRANSFORM, a second CD19 CAR-T, lisocabtagene maraleucel, also beat chemotherapy-plus-transplant as second-line treatment, with a low rate of severe side effects.

For lymphoma that came back within a year, going straight to CAR-T beat the decades-old chemotherapy-then-transplant approach, and later improved survival.

In triple-negative breast cancer that had already been through at least two treatments, the TROP2 antibody-drug conjugate sacituzumab govitecan roughly doubled the time patients lived compared with standard chemotherapy.

The ASCO 2021 guideline is the consensus rulebook for checkpoint-inhibitor toxicity: grade the problem, hold or stop the drug, give steroids early, escalate to other immunosuppressants if they fail, and involve specialists.

Immune cells harvested from a patient's tumour, expanded in the lab and reinfused shrank melanoma in 36% of patients whose disease had progressed on PD-1 blockade, with responses that mostly lasted.

A single infusion of BCMA-directed CAR-T cells produced responses in 97% of patients whose myeloma had failed a median of six prior lines, with two-thirds reaching stringent complete response.

Adding the immunotherapy nivolumab to first-line chemotherapy helped patients with advanced stomach and oesophageal adenocarcinoma live longer, especially when the tumour showed PD-L1, making chemo-immunotherapy the new standard.

Combining the multi-kinase inhibitor lenvatinib with pembrolizumab more than doubled the time to progression compared with sunitinib in advanced clear-cell kidney cancer and improved survival, with 71% of patients responding.

Ten years after England began vaccinating 12- to 13-year-olds with the bivalent HPV vaccine, cervical cancer in that cohort had fallen by 87% and severe precancer (CIN3) by 97%.

FDA oncology leaders argued that the maximum tolerated dose paradigm inherited from chemotherapy produces targeted drugs and immunotherapies dosed far above what is needed, using sotorasib (960 mg vs 240 mg) as the case study, and launched Project Optimus to require dose optimisation before approval.

When oncologists received a geriatric assessment summary with management recommendations for patients aged 70 or over, grade 3-5 toxicity fell from 71% to 51% with no loss in survival.

One year of pembrolizumab after surgery for high-risk kidney cancer reduced relapses by about a third and, in later follow-up, became the first adjuvant treatment to help kidney cancer patients live longer.

In women born with a BRCA1 or BRCA2 mutation whose early breast cancer was high risk, a year of the PARP inhibitor olaparib after standard treatment cut relapses by more than 40% and later improved survival.

A radioactive drug that homes to the PSMA protein on prostate cancer cells helped men with heavily pretreated metastatic prostate cancer live about four months longer, launching radioligand therapy as a mainstream treatment.

After surgery for early-stage EGFR-mutated lung cancer, three years of osimertinib cut recurrences by about 80% and later reduced deaths by half.

In carriers of Lynch syndrome, 600 mg of aspirin a day for about two years reduced colorectal cancer over the following decade (HR 0.65 by intention to treat; 0.56 in those who took it for at least two years).

The third-generation ALK inhibitor lorlatinib kept ALK-positive lung cancer under control for years longer than crizotinib and largely prevented brain metastases; at five years most patients on lorlatinib had still not progressed.

In ELEVATE-TN, acalabrutinib, a second-generation BTK inhibitor, with or without an antibody, cut the risk of progression by 80-90% compared with chlorambucil-obinutuzumab in older or unfit patients with CLL.

Combining the immunotherapy atezolizumab with the anti-blood-vessel antibody bevacizumab helped patients with advanced hepatocellular carcinoma live longer than sorafenib, ending a decade in which nothing had beaten that drug.

For the roughly 5% of metastatic bowel cancers with defective DNA mismatch repair, pembrolizumab alone doubled the time without progression compared with chemotherapy, with far fewer side effects and many long-lasting remissions.

Adding two years of the CDK4/6 inhibitor abemaciclib to standard hormone therapy after surgery cut the risk of the cancer coming back by a quarter in women with node-positive, high-risk disease.

In a Dutch-Belgian trial, CT screening using nodule volume rather than diameter cut lung cancer deaths in men by about a quarter at 10 years, with a far lower false-positive rate than NLST.

Among 1.7 million Swedish girls and women, quadrivalent HPV vaccination cut invasive cervical cancer by 88% when given before age 17 and by about half when given at 17-30.

Adding the BCL-2 inhibitor venetoclax to azacitidine more than doubled remission rates and extended median survival from 9.6 to 14.7 months in unfit AML patients.

In mantle cell lymphoma that had stopped responding to BTK inhibitors, a single CAR-T infusion produced responses in 93% of patients and complete remissions in two-thirds.

An oral FLT3 inhibitor extended survival compared with salvage chemotherapy in relapsed AML with a FLT3 mutation, doubling the remission rate.

A fixed one-year course of two targeted drugs kept CLL under control far longer than chemo-immunotherapy, and most patients had no detectable disease when treatment stopped.

In adults with aggressive lymphoma after at least two prior treatments, tisagenlecleucel produced responses in 52% and complete responses in 40%, most of which lasted.

Women whose HER2-positive breast cancer was still present at surgery after chemotherapy and trastuzumab had half the risk of relapse if their post-surgery treatment was switched to the antibody-drug conjugate T-DM1 instead of continuing trastuzumab.

Pembrolizumab, either alone in PD-L1-positive tumours or combined with chemotherapy, helped patients with recurrent or metastatic head and neck squamous cell cancer live longer than the previous cetuximab-based standard.

Adding the CD38 antibody daratumumab to standard lenalidomide-dexamethasone cut the risk of progression or death by about 44% in older myeloma patients, and later extended survival.

Adding the PARP inhibitor olaparib to bevacizumab after first-line chemotherapy for advanced ovarian cancer roughly doubled the time without progression in women whose tumours had defective DNA repair, but did nothing for those without it.

Swapping bleomycin for the CD30 antibody-drug conjugate brentuximab vedotin modestly improved disease control in advanced Hodgkin lymphoma and, at six years, improved survival.

Across 25 centres, 81% of children and young adults with relapsed or refractory ALL went into remission after a single tisagenlecleucel infusion, and half were still event-free a year later.

Starting with the third-generation EGFR drug osimertinib, rather than saving it for later, kept EGFR-mutated lung cancer under control for almost twice as long and later helped patients live longer.

Adding pembrolizumab to standard chemotherapy roughly halved the risk of death in newly diagnosed non-squamous lung cancer, whatever the PD-L1 level, making chemo-immunotherapy the default first treatment.

In relapsed CLL, a time-limited venetoclax-rituximab course cut progression risk by more than 80% compared with bendamustine-rituximab and later improved survival.

A meta-analysis of 56 accuracy studies and 25 randomised trials found that self-sampled swabs tested with PCR detect precancer as well as clinician-taken samples, and mailing kits to under-screened women roughly doubles participation.

In women with BRCA-mutated advanced ovarian cancer, two years of the PARP inhibitor olaparib after chemotherapy cut the risk of progression or death by 70%, and years later many more were still alive.

Patients on chemotherapy for advanced cancer who reported 12 symptoms weekly through a web tool, with nurse alerts for severe symptoms, lived a median 5 months longer than those receiving usual care.

Across 86 patients with 12 different dMMR cancers, pembrolizumab produced responses in 53% and complete responses in 21%, prompting the first approval of a cancer drug based on a genetic marker rather than tumour site.

Giving the immunotherapy durvalumab for a year after chemoradiotherapy for unresectable stage III lung cancer tripled the time to progression and raised five-year survival from about a third to over 40%.

Giving the hormone-pathway drug abiraterone from the start, alongside standard testosterone suppression, cut deaths by more than a third in men newly diagnosed with high-risk or metastatic prostate cancer.

In lymphoma refractory to chemotherapy, where expected survival was around six months, a single CAR-T infusion produced responses in 82% of patients and long-term remission in about 40%.

An IARC working group reviewed over 1,000 studies and concluded there is sufficient evidence that absence of excess body fat lowers the risk of 13 cancers, up from 5 in the 2002 evaluation.

A CD22 antibody-drug conjugate produced complete remission in 81% of adults with relapsed ALL compared with 29% on chemotherapy, at the cost of liver toxicity in about one in ten.

In women at increased risk, 5 years of tamoxifen reduced breast cancer by 29% over a median 16 years of follow-up, with the benefit continuing long after the pills stopped, but no reduction in breast cancer deaths.

Pembrolizumab shrank tumours in 40% of colorectal and 71% of other cancers with faulty DNA mismatch repair, but in none with intact repair, tying immunotherapy response to mutation burden.

The first sizeable series of CD19 CAR-T therapy showed complete remission in 90% of patients with leukaemia that had failed everything else, with persistent engineered cells and a new, treatable toxicity.

Ruxolitinib shrank the enlarged spleen by more than a third in 42% of myelofibrosis patients versus under 1% on placebo, and halved symptom scores in nearly half.

Nivolumab shrank tumours in roughly a fifth to a quarter of patients with three different advanced cancers, with responses that lasted more than a year and a hint that PD-L1 on the tumour predicted benefit.

Three annual low-dose CT scans reduced lung cancer deaths by a fifth compared with chest X-rays in current and former heavy smokers.

Blocking the immune brake CTLA-4 was the first treatment ever to prolong life in metastatic melanoma, with about one in five patients alive at two years and a new class of autoimmune side effects.

Among 2,482 women with BRCA1 or BRCA2 mutations, removing the ovaries and fallopian tubes was associated with 60% lower all-cause mortality, and no breast cancers occurred in the women who had preventive mastectomy.

Patients newly diagnosed with metastatic lung cancer who saw a palliative care team from diagnosis had better quality of life, less depression, less aggressive end-of-life care and lived a median 2.7 months longer than those receiving oncology care alone.

Adding the oral chemotherapy temozolomide during and after radiotherapy extended median survival in glioblastoma by about two and a half months and more than doubled the number of patients alive at two years; twenty years later it is still the standard.

After following 34,439 male doctors for 50 years, lifelong smokers died on average about 10 years earlier than never-smokers, and stopping at 60, 50, 40 or 30 recovered about 3, 6, 9 or the full 10 years.

Imatinib beat the previous standard by a wide margin in newly diagnosed CML, and the long-term follow-up showed most patients alive at ten years.

In the first human study of imatinib, almost every patient with chronic-phase CML who had failed interferon regained normal blood counts, with mild side effects.

Annual faecal occult blood testing followed by colonoscopy for positives reduced colorectal cancer deaths by 33% over 13 years, the first proof that bowel cancer screening saves lives.

Comparing 649 men with lung cancer to matched hospital controls in London, Doll and Hill found that almost none of the cancer patients were non-smokers and that risk rose steeply with the amount smoked.

In the first randomised trial to beat pembrolizumab directly, a single antibody that blocks both PD-1 and VEGF nearly doubled the time to progression in PD-L1-positive lung cancer, though survival data were still immature.

A vaccine custom-made from each patient's own tumour mutations, given with pembrolizumab after surgery for high-risk melanoma, reduced recurrence by about 44% compared with pembrolizumab alone in a mid-sized randomised trial, the first sign that personalised cancer vaccines can work.

The TROP2-directed antibody-drug conjugate Dato-DXd delayed progression by about two months compared with chemotherapy, but patients did not live longer, which stalled its approval in breast cancer.

In 1,039 Japanese patients with resected colorectal cancer, a positive Signatera test at week 4 carried a tenfold higher recurrence risk, and only ctDNA-positive patients appeared to benefit from adjuvant chemotherapy.

In the first randomised trial of AI in population mammography, AI-supported reading detected 20% more cancers than standard double reading without increasing false positives, and cut screen-reading work by 44%.

In 6,621 adults over 50, the Galleri test flagged a cancer signal in 1.4%, of whom 38% turned out to have cancer; diagnostic work-up took about two months for true positives and over five months for false positives.

Individualised mRNA vaccines encoding up to 20 of each patient's tumour mutations generated strong T-cell responses in half of pancreatic cancer patients after surgery, and those responders had far fewer relapses.

Analysis of 1,644 tumour regions from 421 patients confirmed that subclonal expansions and whole-genome doubling predict relapse, mapped which drivers are selected late, and showed that the metastasising subclone is often a minor population in the primary.

The third hallmarks paper proposed two new hallmarks (unlocking phenotypic plasticity, senescent cells) and two new enabling characteristics (non-mutational epigenetic reprogramming, polymorphic microbiomes), bringing the framework to fourteen elements.

The NHS-Galleri design paper is the protocol for a 140,000-person randomised trial in England testing whether three annual Galleri blood tests reduce late-stage (III-IV) cancer diagnoses, the first MCED trial powered for a stage-shift endpoint.

The first randomised trial of colonoscopy screening found an 18% reduction in bowel cancer incidence after 10 years among those invited, with no significant reduction in bowel cancer deaths; only 42% of invitees actually had the procedure.

Building on the IARC classification of alcohol as a Group 1 carcinogen, this global modelling study attributed 4.1% of all new cancers in 2020 to drinking, three-quarters of them in men and the largest numbers in oesophageal, liver and breast cancer.

DeepMind's neural network predicted protein structures at CASP14 with a median backbone error of under 1 angstrom, comparable to experimental methods, and the team released predicted structures for essentially every human protein within a year.

An eight-year effort to repeat 50 experiments from 23 high-impact cancer biology papers found that replication effect sizes were on average 85% smaller than the originals, and fewer than half of the effects replicated by most criteria.

The first prospective interventional study of a multi-analyte blood test (CancerSEEK) in asymptomatic people: 26 cancers were first detected by the blood test, most of them localised, and the test roughly doubled the number of cancers caught by standard screening.

A vaccine of three frameshift peptides shared by almost all microsatellite-unstable tumours was safe and induced immune responses in every patient with MSI colorectal cancer, opening the path to cancer interception vaccines for Lynch syndrome.

Off-the-shelf CAR-T cells made from a healthy donor, gene-edited to avoid rejection and graft-versus-host disease, produced remission in 14 of 21 patients with relapsed ALL.

Pooling 13 studies of 8,883 patients, 56% had no trial available at their site and a further 22% were ineligible for the trials that existed; among patients actually offered a trial, roughly half enrolled, so overall participation was about 8%.

The capstone of The Cancer Genome Atlas integrated DNA, RNA, protein and methylation data on about 10,000 tumours, showing that cell of origin dominates molecular classification but that some cancers regroup across organs.

Genome-scale RNAi screens across 501 cancer cell lines, analysed with the DEMETER algorithm to remove off-target noise, identified 769 genes on which subsets of cancers depend and showed that most dependencies can be predicted from the cell's molecular features.

Multi-region sequencing of 327 regions from the first 100 TRACERx lung cancers showed that most driver mutations are early and shared while copy-number chaos continues to evolve, and that tumours with high copy-number heterogeneity were nearly five times more likely to relapse or kill the patient.

Deep sequencing of 234 tiny biopsies of normal eyelid skin from four people found thousands of clones carrying cancer driver mutations, with about a fifth to a third of cells carrying NOTCH1 mutations, showing that driver mutations are common in healthy tissue.

A systematic review of 36 trial-level meta-analyses found that in over half the surrogate endpoint (response rate, progression-free survival) had a low correlation with overall survival, and only about a quarter showed a strong correlation.

In a Spanish cardiovascular prevention trial, women assigned to a Mediterranean diet supplemented with extra-virgin olive oil had about two-thirds fewer invasive breast cancers than a low-fat control group, although the number of cases was small.

Blood DNA from 17,182 people showed that clones carrying leukaemia-associated mutations (mostly DNMT3A, TET2, ASXL1) are present in about 10% of people over 70, raising the risk of blood cancer 11-fold and, unexpectedly, of heart attack and stroke.

Synthesising the first generation of cancer genome sequencing, Vogelstein and colleagues concluded that about 140 genes can drive cancer when mutated, that a typical solid tumour has 2 to 8 driver mutations among tens of passengers, and that all drivers act through a dozen signalling pathways.

Using disulfide-tethering screens, Shokat's laboratory found small molecules that bind covalently to the mutant cysteine of KRAS G12C in a previously unknown pocket beneath switch II, locking the oncoprotein in its inactive GDP-bound state.

Sequencing multiple regions of four kidney cancers and their metastases showed that around two-thirds of mutations were not shared across the whole tumour, that good- and poor-prognosis gene signatures coexisted in the same tumour, and that different regions had independently hit the same genes.

A decade after Honjo's group cloned PD-1 (1992), this study showed that tumour cells expressing PD-L1 resist killing by cytotoxic T cells in mice, and that anti-PD-L1 antibody or PD-1 deficiency restores tumour rejection, the foundation of PD-1/PD-L1 therapy.

Crews and Deshaies built a two-headed molecule linking a ligand for the target protein MetAP-2 to a peptide recognised by an E3 ubiquitin ligase, and showed it induced ubiquitination and degradation of the target, founding targeted protein degradation.

Hanahan and Weinberg distilled decades of cancer biology into six acquired capabilities shared by all cancers, giving the field a common organising framework that has been cited more than any other cancer paper.

Injecting mice with an antibody that blocks the T-cell inhibitory receptor CTLA-4 caused established colon carcinomas and fibrosarcomas to be rejected and protected against re-challenge, the experiment that founded checkpoint immunotherapy.