Part 2
If you have read Part 1 of this editorial, you may recall that in 1936, Jorge Luis Borges analyzed the various translations of One Thousand and One Nights, a collection of Middle Eastern and Indian stories sometimes referred to as Arabian Nights. I found myself thinking of his conclusions as I considered the thoracic oncology research presented at the 2026 American Society of Clinical Oncology (ASCO) Annual Meeting: That when examining the Nights, there is no definitive text—only successive versions.
I see a similar trend playing out in lung cancer research. As discussed in Part 1, the trials presented during the plenary session (LIBRETTO-432 and HARMONi-6) and the RET and ALK target families gave my conceit its first tests. The rest of the meeting continued to fuel it.
The latest EGFR research arrived with three different versions, splitting the resistance problem into solvable thirds. CHRYSALIS-2 reported an updated overall survival (OS) of 41 months for first-line amivantamab plus lazertinib in patients with atypical EGFR-mutant non-small cell lung cancer (NSCLC), a subgroup for which afatinib, the previously approved option, had delivered roughly 19 months across cross-trial comparisons; 46% of treated patients were alive at 4 years.1
Silevertinib (BDTX-1535), a fourth-generation EGFR TKI, treated the de novo non-classical mutations that osimertinib leaves behind: It yielded a 60% overall response rate (ORR), 56% specifically in patients with PACC mutations. The data showed 86% intracranial response with 81% of patients exhibiting ctDNA EGFR clearance.2 DZD6008 treated acquired resistance, going after C797X, the dominant on-target resistance mutation after osimertinib, with a 60% ORR at the higher recommended phase II dose and a notably mild safety profile.3
Atypical, non-classical, and post-osimertinib resistance: Three translators “reading three different EGFR texts.”
KRAS arrived as a family. The allele, once considered undruggable, now has multiple potential treatment pathways. Divarasib plus pembrolizumab in first-line KRAS G12C-positive disease produced a 73% ORR and 19.3-month progression-free survival (PFS) as a chemotherapy-free option, with activity even in PD-L1-negative tumors, but at the cost of grade 3-4 ALT elevation in 20% of patients.4
Elisrasib, a next-generation GDP-bound G12C inhibitor, achieved 78% ORR as monotherapy in the first line—the first such evidence for a G12C inhibitor—and 81% in combination with pembrolizumab, and a far gentler hepatic profile with grade 3-4 ALT elevations of 7.7%.5 TSN1611, an oral G12D inhibitor, posted a 50% ORR in 20 response-evaluable patients with NSCLC with intracranial responses and a 9.4% incidence of grade 3 toxicity.6
G12D is the most common KRAS allele and the dominant driver in pancreatic and colorectal cancer; its first credible step into the druggable column is one of the structural milestones of the meeting. The hepatic differential between divarasib and elisrasib is itself a sub-translation: liver liability appears to be molecule-specific, not class-wide.
The pattern is Borges’s, exactly. The target is the text, and each translator brings a separate story of durability, CNS penetrance, tolerability, chemotherapy-free design, and sequencing. And each version is unmistakably itself.
Pierre Menard, the Author of Adjuvant Erlotinib
In 1939, Borges wrote about the fictional French symbolist Pierre Menard, who set himself the task of rewriting Don Quixote. Not from memory, not as an adaptation, but word for word in the original Spanish. Borges argued that Menard’s Quixote, identical to Cervantes’ original in every syllable, is a different work because it was written by a different man in a different century in a different language. The text is the same. The context has rewritten it.7
Three other lung trials presented at ASCO 2026 read this way.
Alliance A081105 (Govindan) tested adjuvant erlotinib versus observation in resected EGFR-mutant NSCLC. After mature follow-up, disease-free survival (DFS) improved, 68 versus 50 months. OS did not. Five-year OS was 78.6% versus 77.9%.8 Enrollment had been stopped early once adjuvant osimertinib proved itself in ADAURA. The trial finished while the therapeutic world moved past its protocol. The erlotinib of 2008 and osimertinib, its replacement since the late 2010s, are not the same drugs, even if the agents are very similar with respect to molecular construct and mechanism of action.
EA5142 (Chaft) tested adjuvant nivolumab after chemotherapy in resected EGFR/ALK wild-type NSCLC. At 72.6 months of follow-up, the DFS hazard ratio was 0.97, p = 0.78, with no benefit seen in any subset, including PD-L1 ≥ 50% [HR 0.86, p = 0.43].9 Adjuvant-only PD-1, untethered from the perioperative design that subsequent trials have validated, does not deliver the neoantigen-priming that the in situ tumor enables. The translation failed not because the drug failed, but rather because the study design had been retranslated around its intent.
The Tata Memorial brain-metastasis trial tested upfront cranial radiotherapy (XRT) versus deferred XRT in asymptomatic brain metastases in EGFR/ALK-positive NSCLC. Cranial radiotherapy slashed intracranial progression, 21.7% versus 50% at 2 years [HR 0.35], but there was no improvement in overall survival or progression-free survival, and a 25% incidence of radiation necrosis.10 In the pre-TKI era, this would have been considered a positive trial. In the modern era, the standard for success has become more exacting.
These studies do not necessarily feature failed drugs or strategies. But these are versions in which the original was rewritten while we were still reading.
The Second Translation
The theme set by ASCO Immediate Past President Eric J. Small, MD, had a second register that the meeting could not avoid. The most striking data presented in 2026 underscored a recurring caveat—highly positive, China-only trials—and this caveat is not just a footnote. It potentially poses a major translation problem.
HARMONi-6 against tislelizumab is the clean case. OptiTROP-Lung05 is another: Sacituzumab tirumotecan plus pembrolizumab versus pembrolizumab alone in first-line PD-L1-positive NSCLC reduced the risk of progression or death by 65% and the risk of death by 45% in 413 patients. It is the first phase III win for an antibody-drug conjugate plus immunotherapy in this space.11 However, the comparator is fair only for patients with PD-L1 ≥ 50% NSCLC.
For the 1% to 49% group, where more tumors reside, the global standard since KEYNOTE-189 and KEYNOTE-407 has been histology-appropriate chemotherapy plus pembrolizumab, and the trial failed to address this flaw. The KRAS data, the fourth-generation EGFR data, and the PD-1/VEGF bispecific class read the same way: Striking signals observed in unique populations that the rest of the world will want to retranslate before adoption.
The harder translation is the second one. To matter at scale, a drug must travel not only from one population’s data to another’s but from a genome that can now be read anywhere to a treatment a patient can actually obtain anywhere. The WHO Essential Medicines List has steadily expanded in oncology during the past decade. Unfortunately, the gap between listing and affordability has not closed at the same pace.
The Access to Oncology Medications Coalition’s first four years—with more than a million doses of quality-assured cancer medicines delivered across 46 countries and 19 cancer types—are evidence that the gap can be narrowed by infrastructure, not only by lower prices. But the molecular vocabulary of 2026 was written largely in centers where access is assumed. The patients who will most need the translations of LIBRETTO-432, HARMONi-6, AcceleRET-Lung, and the rest of these successes are the ones least likely to read them.
The Nights
The Nights, Borges reminded us, were never a stable text. Each translator left fingerprints, and the compendium grew through them, not despite them.
This has been a year of retranslations. CROWN at 7 years confirmed the version we already had. HARMONi-6 opened a new one, in a single language, and may or may not survive its second reading. ARCHER and the KRAS family are the first lines of versions still being written. ALCHEMIST is the version we no longer read because the original was rewritten while we held the text.
Again, as Borges wrote in his essay on the Greek translators of Homer with a sentence that has aged into a small monument: El concepto de texto definitivo no corresponde sino a la religión o al cansancio. The concept of a definitive text belongs only to religion or to fatigue.
References
- 1. Neal JW, et al. Overall survival of first-line amivantamab plus lazertinib in atypical EGFR-mutated advanced NSCLC: updated CHRYSALIS-2. J Clin Oncol. 2026;44(suppl):8501.
- 2. Rotow J, et al. Silevertinib (BDTX-1535) in non-classical EGFR-mutated NSCLC. J Clin Oncol. 2026;44(suppl):8519.
- 3. Wang Y, et al. DZD6008, a fourth-generation EGFR TKI targeting C797X-resistant NSCLC. J Clin Oncol. 2026;44(suppl):8520.
- 4. Skoulidis F, et al. Krascendo-170: divarasib plus pembrolizumab in first-line KRAS G12C-positive NSCLC. J Clin Oncol. 2026;44(suppl):8510.
- 5. Lu S, et al. Elisrasib, a next-generation GDP-bound KRAS G12C inhibitor, in first-line G12C-positive NSCLC. J Clin Oncol. 2026;44(suppl):8511.
- 6. Yu HA, et al. TSN1611, an oral KRAS G12D inhibitor, in advanced solid tumors. J Clin Oncol. 2026;44(suppl):8516.
- 7. Borges JL. Pierre Menard, autor del Quijote. In: Ficciones. Buenos Aires: Sur; 1944.
- 8. Govindan R, et al. Alliance A081105: adjuvant erlotinib versus observation in resected EGFR-mutated NSCLC – final overall survival. J Clin Oncol. 2026;44(suppl):8001.
- 9. Chaft JE, et al. EA5142 (ANVIL): adjuvant nivolumab after chemotherapy in resected NSCLC. J Clin Oncol. 2026;44(suppl):8000.
- 10. Tibdewal AR, et al. Upfront cranial radiotherapy versus deferral in asymptomatic brain metastases from oncogene-mutated NSCLC: phase 3 randomized trial. J Clin Oncol. 2026;44(suppl):8624.
- 11. Zhou C, et al. OptiTROP-Lung05: sacituzumab tirumotecan plus pembrolizumab versus pembrolizumab in first-line PD-L1-positive advanced NSCLC. J Clin Oncol. 2026;44(suppl):8506.
