Chemotherapy became the standard treatment for lung cancer in the twentieth century1. But in the past 15 years, there has been a drive to improve outcomes for people with this still-deadly disease, either through therapies that target enzymes encoded by genes harbouring cancer-driving mutations, or through immunotherapies, which activate the body’s immune system to target tumours. Writing in The New England Journal of Medicine, two groups2,3 provide evidence that supports the use of immunotherapies to treat non-small-cell lung cancer (NSCLC) at different stages of the disease.
Tumour cells evade destruction by activating signals known as immune checkpoints, which deactivate immune cells called T cells4. Two immune checkpoints are the proteins cytotoxic T lymphocyte antigen 4 (CTLA-4) and programmed cell death 1 (PD-1), which are expressed by T cells themselves. Another, programmed cell death ligand 1 (PD-L1), is produced by tumour cells (Fig. 1).
Antibodies that interact with these proteins to prevent their normal activity, and so reawaken the immune system, are now used to treat metastatic NSCLC — the stage at which the cancer has spread. Antibodies that bind PD-1 or PD-L1 are sometimes successful in patients who have had treatments such as chemotherapy, but whose cancer has nonetheless progressed5. Alternatively, the anti-PD-1 antibody pembrolizumab can be used as a first-line treatment for metastatic NSCLC when the percentage of tumour cells that express PD-L1 is high — these patients respond better to immunotherapy than to chemotherapy6.
If such immune-checkpoint-targeted antibodies (ICTs) can improve outcomes for metastatic NSCLC, could they also help to tackle early-stage disease? In the first of the current papers, Forde et al.2 carried out a pilot study to investigate whether the anti-PD-1 ICT nivolumab could be used to shrink tumours before surgery, which is a standard treatment for most cases of early-stage NSCLC.
The authors treated 21 patients with 2 doses of nivolumab 2 weeks apart, starting 4 weeks before the planned surgery date. They showed that surgery did not need to be delayed (for example, because of an adverse event with nivolumab) for any patient. The researchers anticipated that four weeks would not be enough time for the reactivated immune system to significantly shrink the tumour. Indeed, imaging revealed significant shrinkage in tumours in only two patients before surgery. However, examination of the surgically removed tumours revealed that 45% had undergone a major response to the ICT — less than 10% of the tumour cells remained alive. ICTs, unlike chemotherapy, cause inflammation and scar-tissue formation in tumours, and can therefore sometimes cause tumour growth. However, the researchers found that even two tumours that showed such growth had undergone a strong pathological response.
This level of efficacy is impressive, but needs to be further investigated by in-depth examination of the tumour specimens collected. Moreover, the optimal duration of nivolumab treatment remains to be determined, because delayed responses to ICTs can occur7. Of note, one of the patients enrolled experienced severe acute toxicity to nivolumab. Possible long-term side effects also need to be considered, because Forde et al. followed their patients for only a median of 12 months after surgery. Phase III trials are therefore now essential.
Forde and colleagues found that the number of mutations in each tumour’s genome correlated with whether that tumour had a major pathological response to the ICT. And another study8 has found that a high number of mutations in a tumour — the tumour mutational burden (TMB) — is associated with an improved outcome in patients treated with a combination of ICTs. So, can TMB be used to predict which patients with advanced NSCLC would benefit from immunotherapy? In the second of the current papers, Hellmann et al.3 investigated this possibility, as part of a phase III trial called CheckMate 227.
The authors selected patients who had untreated, advanced-stage NSCLC. They assigned patients to one of four treatment groups — nivolumab; chemotherapy plus nivolumab; chemotherapy; or nivolumab plus the anti-CTLA4 antibody ipilimumab. For the last two groups, they analysed how a high TMB of at least 10 mutations per megabase of DNA affected progression-free survival — the time before the tumour begins to grow or spread once more.
Hellmann and colleagues found that median progression-free survival for patients who had a high TMB was 7.2 months for nivolumab plus ipilimumab, compared with 5.5 months for chemotherapy. After one year, there was no progression in 42.6% of patients who received the combination immunotherapy, compared with 13.2% of those who received chemotherapy alone. However, people with a low TMB did not benefit from combination ICT. These results are remarkable — the first positive results for a combination ICT predicted using TMB.
What will the impact of these findings be, in an already crowded treatment field? One issue is the feasibility of using TMB to find suitable patients. Out of the 1,739 patients enrolled by Hellmann et al., only 1,004 could be evaluated for TMB, mainly because the amount of tissue available or the quality of the DNA extracted was inadequate. The analysis of progression-free survival in patients with a high TMB involved only 299 people. Furthermore, there are multiple tests for TMB that analyse different genomic regions, and multiple cut-offs for high TMB classification — no cross-test validation exists. Other drawbacks of TMB analysis are the cost; the amount of tissue needed; and the fact that the analysis takes about two weeks. Finally, some of the tumours that had high TMB in Hellmann and colleagues’ study also had high levels of PD-L1 expression. In these cases, single-agent immunotherapy is effective, and is less toxic than the combination treatment6.
It should be noted that two recently reported phase III trials9,10 found that the combination of chemotherapy and an ICT is superior to chemotherapy, whatever the level of PD-L1 expression (and possibly whatever the TMB). It is expected that these studies will establish a new standard of care. It will be hard for ICT combinations to compete.
ICTs can also have a dark side. One study11 found that 14% of people with NSCLC who were treated with ICTs developed hyperprogressive disease — an increase in tumour growth rate compared to the rate when patients were given a previous treatment. Hyperprogression is associated with poor survival rates. In the first few months of Hellmann and colleagues’ trial, progression-free survival was higher for the chemotherapy arm than for the combination ICT, possibly suggestive of hyperprogression in some patients on the combination ICT. Whether such a pattern might arise in early-stage NSCLC, such as that examined by Forde et al., should also be carefully monitored. Finally, ICTs could harm our health-care systems through their cost — more than US$10,000 per month for a combination treatment12. We will need to monitor whether such a cost is sustainable.
Nature 558, 376-377 (2018)