We've covered the potential value of systemic therapy for early stage NSCLC in a wide range of posts and podcasts, and to summarize what we've learned in a sentence, it's basically that chemotherapy can significantly increase progression-free survival (PFS) and overall survival (OS) in patients who have undergone curative surgery for stage I-III NSCLC, but the benefit is far more convincing in patients with a high enough risk to justify the potential adverse effects of chemotherapy.  In fact, in patients with node-negative cancers that are smaller than about 4 cm, the evidence isn't very good that chemotherapy improves outcomes, and there is a strong suggestion from the limited available evidence that it may be net harmful.  This makes sense to me: chemotherapy represents a fixed negative effect (both quality of life and risk of adverse effects) whether a patient has a high risk of recurrence or a low risk of recurrence and can reduce the chance of the cancer recurring by a certain proportion. If the risk of recurrence is high (such as in someone with stage II or III resected NSCLC), chemo can reduce that risk quite a bit, so the net effect is very positive: big anticipated benefit exceeds small risk.  However, in patients with smaller node-negative cancers, the magnitude of benefit is going to be very low because the risk of recurrence is too low for the chemo to have much absolute effect: small risk exceeds even smaller anticipated benefit.

In recent years, we have generally focused on post-operative, or adjuvant chemotherapy, because the majority of positive trials for early stage patients have used an adjuvant strategy.  However, going back in time ten years, both pre-operative (neoadjuvant) and post-operative chemotherapy were investigational approaches, and they each have their advantages and disadvantages.  Post-operative chemotherapy has the benefit of being able to make treatment recommendations based on the most accurate staging information (from surgery) and provides the opportunity to do the most pivotal treatment immediately, but many patients are simply not able to consider chemotherapy within the first couple of months after surgery, and/or they need to abort treatment before the intended therapy has been delivered.  Pre-operative chemotherapy provides the earliest opportunity to treat potential micrometastic disease, should improve the probability that treatment will be delivered as planned (because few people will decide to abandon surgery), and it gives the chance to get feedback on how effectively the systemic therapy shrunk the cancer.  However, it also entails a small but real chance that the person's cancer will grow and even potentially no longer be able to be resected, and possibly that chemo could increase the risk of surgical and post-surgical complications.   Pre-operative therapy may also have an advantage of allowing a patient to receive a less extensive surgery after a good response to initial systemic treatment.

So it's fair to say that there is a good rationale to test both of these strategies, and studies have been pursued with surgery followed by randomization to chemo or observation, immediate randomization to chemotherapy followed by surgery or immediate surgery alone, and even one trial that randomized patients to pre-operative chemotherapy followed by surgery, surgery followed by chemotherapy, or neither pre-operative nor post-operative therapy and just surgery alone.   By 2003 and 2004, several trials of post-operative chemotherapy were reported as positive, and this led to early closure of several of the important pre-operative chemotherapy trials, since the emerging picture was that chemotherapy provided a benefit that made randomization of patients to a surgery alone arm unethical.

It is in this context that we can now review the significance of the newly published Chemotherapy in Early Stages Trial (ChEST), which was conducted at 45 centers in 15 countries in Europe and attempted to assess the benefit of three cycles of cisplatin/gemcitabine before surgery, compared with surgery alone.  The trial was designed with an intent to enroll 712 patients with stage IB to stage IIIA NSCLC (stage IIIA only if they had no mediastinal nodal involvement, so T3N1) and was looking to detect a 20% improvement in PFS as the primary endpoint, but from 2000 to 2004 only 270 patients were enrolled (129 randomized to pre-op chemotherapy, and 141 randomized to surgery alone).  It closed early in light of the mounting evidence supporting a role for chemotherapy.

When we look at who specifically was on the trial, it's not unusual for a European study: over 80% were men, about 98% Caucasian, and with squamous histology as the most common (about 40%).  Importantly, the staging was overwhelmingly dominated by stage IB (primary tumor larger than 3 cm, with no nodal involvement) or stage IIB (most patients with a tumor larger than 3 cm, and with N1 nodal involvement), with just a smattering of patients with another stage.

When we look at how patients did with chemotherapy, 127 of the 129 assigned to it received at least some chemotherapy, and 109 (84%) received all three cycles.  This is a higher proportion than we typically see in adjuvant therapy trials, where only about 60-70% of patients typically get through the intended regimen (and we don't know how many people aren't even considered for the trial because they still having a hard time recovering from surgery).   Repeat imaging after chemotherapy showed that 35% had a significant response, and 3% had a complete response (no evidence of disease).  Most of the remainder had what would be considered stable disease, with progressive disease (significant growth or new lesions) only seen in 5.5% of patients.  Arguably related to this, fewer patients on the neoadjuvant chemotherapy arm underwent the large surgery of a pneumonectomy (17% vs. 25%) and more underwent a lobectomy (70% vs. 60%), so this potential benefit of pre-operative surgery may have been realized. A slightly higher proportion of patients achieved a complete resection ("we got it all out") if they underwent neoadjuvant chemotherapy: 88% vs. 84%.

With regard to side effects, the well-studied regimen of cisplatin/gemcitabine was associated with the side effects you'd expect: drops in blood counts were common, some nausea and vomiting that was rarely severe, and importantly there was no increase in mortality after surgery (3% with pre-operative chemotherapy, vs. 4% with surgery alone).

The big question is how the two groups did in terms of PFS and OS, and despite the small number of patients enrolled, the outcomes with neoadjuvant chemotherapy was still convincingly if not always statistically significantly better with neoadjuvant chemotherapy.  Median PFS was 4.0 years with neoadjuvant chemotherapy, compared with 2.9 years with surgery alone (log rank P=0.09), with a 5% absolute difference in PFS three years after enrollment favoring chemotherapy (52.9% vs. 47.9%).

Differences in OS were statistically significantly better with chemotherapy, with a median OS of 7.8 years vs. 4.8 years (log rank P=0.04), with an absolute difference in survival of 8% at three years out from enrollment (67.6% vs. 59..8%).

(Click on image to enlarge)

It's important to remember that these median numbers don't mean "people should expect to live 8 years" or anything like that: this is a median number where half of the patients had died by that time, and if more patients are cured, the median survival will improve.  The goal here is primarily to improve the proportion of people being completely cured of their lung cancer.

But what was especially interesting was the breakdown of results by whether patients had lower risk or higher risk disease. Specifically, enrollment was nearly evenly split between patients who were stage IB/IIA (mostly IB) and patients who were stage IIB/IIIA (mostly IIB); when looking at these groups separately, the benefit of neoadjuvant chemotherapy was remarkably greater in the patients in the higher risk group. Patients with stage IIB/IIIA NSCLC who received pre-operative chemo had a median PFS of 4.0 years, compared with 1.1 years with surgery alone; in this group, the proportion of patients who were without disease progression at three years was 55.4% vs. 36.1%, favoring the chemotherapy arm -- a difference that is somewhere between impressive and astonishing.

Overall survival followed the same pattern, with benefit from neoadjuvant chemotherapy overwhelmingly more notable in the higher risk patients.   Compared with no real difference in the stage IB/IIA patients, median OS in the higher risk stage IIB/IIIA patients was 2.1 years with surgery vs. "not yet reached/not evaluable" with chemotherapy.  The "hazard ratio" of relative improvement in OS for stage IIB/IIIA patients was 0.42, meaning that the chance of a patient in this group who received chemotherapy would be alive was more than twice as great at a given time point than a patient who underwent surgery alone.

So what can we make of these results? The editorial that accompanied this article by Dr. Gary Strauss from Tufts, who led the CALGB 9633 trial of adjuvant carboplatin/Taxol (paclitaxel) for patients with stage IB NSCLC, highlights the shortcomings of this trial: the fact that the surgery alone arm had a higher proportion of men (who tend to do a little worse than women) and a median age 2 years older than the neoadjuvant chemotherapy arm (again, older patients have a tendency to do a little worse than younger patients) as a potential confounding explanation for these results, which he also stated were so remarkable that they weren't easy to believe.  He also made the point that differences in the benefit of chemotherapy depending on the degree of risk of cancer recurrence haven't been noted in the setting of adjuvant therapy.  Frankly, I don't know where he's been the last 5 years, because that has been a very clear and common theme I've summarized, based on several post-operative chemotherapy trials.   MANY of the adjuvant therapy trials have shown that the benefit of chemotherapy was seen only in stage II or III patients, and only in stage IB NSCLC patients with larger (generally >4 cm) tumors.

My overall impression is that these results are remarkable, with a greater difference than I would expect to see, and that the absolute numbers may not be plausible compared with what we've seen previously (particularly the PFS and OS differences in stage IIB/IIIA patients), but these results are very consistent with the same conclusions I'd drawn from many other studies in early stage NSCLC patients with potentially resectable disease.  Specifically, here are what I would say the data support:

1) chemotherapy added to surgery can improve the PFS and OS for patients fit enough to take 3-4 cycles of platinum-based chemotherapy, with the best evidence for cisplatin

2) The benefit of chemotherapy is disproportionately greater in patients with the highest risk of recurrence.  It may well be net harmful for people with lower risk of recurrence and/or who are less fit for the rigors of chemotherapy.

3) These effects are overall quite comparable for pre-operative vs. post-operative chemotherapy.  Pre-operative chemotherapy may provide the additional benefit that more patients are able to start and to complete chemotherapy, and that tumor shrinkage may allow for a less extensive surgery.

However, at this point, adjuvant therapy has the upper hand and isn't likely to be displaced by neoadjuvant chemotherapy as a standard of care.  First, there are more statistically significantly positive studies supporting adjuvant therapy.  This trial may be considered "positive" in some senses, and in my mind it certainly supports pre-operative chemotherapy for patients with higher risk resectable NSCLC, but it only enrolled a small fraction of the intended trial population and failed to show a statistically significant improvement in the entire trial population in its primary endpoint of PFS.  Other neoadjuvant therapy trials have largely shown moderately favorable results that I'd consider comparable to the benefits seen in the adjuvant setting, but these trials have been largely underpowered, and the fact that many have used carboplatin-based rather than cisplatin-based chemotherapy may have contributed to less impressive benefits than would be seen with cisplatin.  But looking collectively at these results, I'd say that they all support the same theme, and there may even be some particular factors that recommend pre-operative chemotherapy.

Another factor is that adjuvant therapy is more appealing to thoracic surgeons and most patients, compared with a neoadjuvant chemotherapy strategy.  Surgeons can book a patient for surgery in the next 1-2 weeks after meeting them for the first time, compared with sending them to the medical oncologist and waiting for 3 months to see them again, with a small minority progressing or experiencing side effects that keep them from undergoing the surgery they would have been candidates for at the initial visit (though the few patients who progress through neoadjuvant chemotherapy are also almost certainly very unlikely to be long term survivors of surgery alone).  Many patients express that they "just want it out" and are uneasy with pursuing anything that could delay that.

So while I am impressed by these imperfect but very provocative results of the ChEST study, and they lead me to believe that neoadjuvant therapy is a comparable and perhaps even superior choice vs. adjuvant chemotherapy, I think that it may amount to "too little, too late" for this concept in terms of becoming a widely practiced strategy, even if it might well actually be the better one.