A 42-year-old man underwent resection for a Stage II T3N0 colonic tumor followed by postoperative adjuvant therapy with 5-FU/LV. Three years post diagnosis, surveillance CT revealed a single 3-cm hepatic lesion abutting a major artery in segment VIII of the right lobe. PET showed no other liver involvement and no extrahepatic disease. He received six cycles of preoperative FOLFOX/bevacizumab with inadequate response for resection and switched to four cycles of FOLFIRI/cetuximab (after primary tumor histology was determined to be K-ras wild type). He experienced a dramatic response and underwent an R0 resection showing no viable tumor in the resected specimen.
Case Discussion:
DR HALLER This was a 42-year-old man who was diagnosed three years earlier with a Stage II, T3N0 colon cancer. At the time of resection, only nine nodes were examined, which is unfortunately the average in the United States despite the recommendation that 11 or 12 nodes should be analyzed to properly stage this disease. He was only 39 years old at the time of diagnosis, so we considered Lynch syndrome when we worked up the case.
My general rule of thumb is if someone has a Stage II tumor with such a good prognosis that I don’t even consider chemotherapy, that I also don’t conduct surveillance except for colonoscopy. In other words, if I don’t think they’re going to relapse, then I shouldn’t be looking for relapse.
However, this patient was young, and although he was staged N0, only nine nodes were examined, so postoperatively we followed him with CEAs every three months and CT scans of the abdomen and pelvis every six and treated him with adjuvant chemotherapy. If I remember correctly, he didn’t want extensive chemotherapy, so he received 5-FU/leucovorin.
Other than the inadequate nodal sampling, this patient had no other high-risk factors. I knew the MOSAIC data were coming out at that point and in the general Stage II population, the survival at six years had a p-value of 0.996. So you can intuit that potentially in some patients who have poorly differentiated tumors, perforation, obstruction, lymphovascular invasion, few nodes taken or prognoses as bad as Stage III patients, that FOLFOX may be considered.
However, I believe there’s also the option for the patients at very good risk of not administering adjuvant therapy or an intermediate choice of 5-FU, because we know in unselected patients, 5-FU does provide approximately a four percent survival benefit. But that brings the bar up so high that the incremental gain from oxaliplatin was not visible in MOSAIC. This year, we will see the survival data from the MOSAIC trial and the PETTAC-3 data being released, and we will be publishing them together.
Two-and-a-half-years after adjuvant chemotherapy, a CT scan showed a 3-cm hepatic lesion in segment VIII of the right lobe, abutting the major artery from the aorta. A PET scan showed no other lesions. These cases require an experienced radiologist and good imaging. You can’t rely on CT from a PET/CT. In this case, the radiologist could not see a normal rim of tissue adjacent to the vessel, so the patient could not be taken immediately to the operating room.
Not many years ago, even in major centers, this patient would have been considered incurable. Unfortunately in many centers now, patients who should be are not referred for consideration of surgery. Today, in a patient like this, if the surgeon said “No” to surgery, then I might send the images to a more experienced surgeon. Not all surgeons have equal experience — I tend to send them to Steve, because he’s such an easy guy to deal with and he’s happy to do it. He has a good team and if the patient eventually goes to surgery, he’s good to them.
In this case, our surgeon and we felt that other than the location of his tumor, in any other circumstances this patient would have a fairly high cure rate of perhaps over 50 percent, given the length of time between his original surgery and relapse and the fact that only one lesion was present. So we treated him with FOLFOX/bevacizumab, as we typically do in the front-line setting.
The patient received six cycles of treatment and the lesion decreased in size, but the surgeon felt it was still not resectable. We could have continued the same regimen, but it’s generally true that maximal response is attained before six cycles and we didn’t feel additional therapy would result in conversion. We elected to change to FOLFIRI, which is a good regimen for advanced disease, and because his primary tumor was wild-type K-ras, we treated with cetuximab. This is a stable mutation, so we didn’t have to biopsy the metastasis in this case. If it had not been a wild type, then we would have administered FOLFIRI only.
The patient was restaged after four cycles and his tumor had decreased dramatically. Based on the results of the CRYSTAL and OPUS trials, this was not unpredictable. The OPUS trial data were published in the Journal of Clinical Oncology on December 29, 2008, and the response rate was dramatically different with cetuximab — not in the whole population, but in the patients who had wild-type K-ras tumors.
Sometimes when treating metastatic disease, response is not so important. If patients are not symptomatic, not going to surgery and not in pain, control of the disease is important, but you will not make them feel better by using extra therapy. However, in this patient due to the location of the metastasis, response was important, as it is in symptomatic patients.
It was felt that radiographically some normal tissue could be seen, so he was taken to the operating room. The surgeon was experienced and he was able to perform an R0 resection. No viable tumor was present in the resected specimen, which happens, so this patient had a pathologic complete response.
The pathology told us two things. First of all, the R0 resection is imperative for curability, so that was one good prognostic factor. The other is that the tumor response — zero viable tumor — was prognostic and predicted a better outcome. We elected not to administer additional therapy at that time, based on the pathology. We felt we had up-fronted all his systemic treatment with the 10 cycles of chemotherapy, including FOLFOX/bevacizumab and FOLFIRI/cetuximab.
Based on the data from the PACCE trial published in the Journal of Clinical Oncology, Charles Blanke wrote a nice editorial on dual biologics and the evolving data. One data set showed a trend toward a worse outcome in PFS and response when you add cetuximab to chemotherapy for patients with mutated K-ras tumors.
Patients with wild-type versus mutant disease always have better prognosis independent of the chemotherapy we administer. That’s another positive predictor of outcome in this patient. I believe that the data are strong enough to divulge that to a patient, because people need to hear good news, especially when you’re 42 years old and you have metastatic colon cancer and underwent a liver resection.
The FDA Oncology Drug Advisory Committee met about a month ago, and they haven’t made a decision regarding K-ras testing. Apparently, some people feel we need to do formal prospective trials, since all of the data we have are retrospective and they sometimes come from only subgroups, like in the CRYSTAL trial, and not the whole patient population.
I studied results of the EPIC trial, which evaluated second-line FOLFIRI with or without cetuximab in 5-FU/oxaliplatin failures. It’s only 300 out of 1,200 patients and only the subsets that were treated in the United States, so many biases can be built into it, but we don’t have the resources for large prospective trials. I don’t believe anyone feels they want to conduct a study of formal predictive testing — that is, to treat both mutant and nonmutants disease and go through 2,400 more patients. It’s not that we only have data from one study. We have the panitumumab study, the cetuximab NCIC study, the CRYSTAL trial, the OPUS trial and every study shows the same effect.
It may not be great science in some respects, but we always conduct imperfect studies. To me, when we have a series of 15 or 20 large, somewhat flawed studies, all showing exactly the same effect on a forest plot, it’s not problematic. Additionally, when a biologic plausibility also explains why it should be true — which exists in this case — then it’s even stronger. If it were an oddball observation and no good biologist, not even Lee Ellis, could figure out why this could be true, then you might make a different decision.
We have to be careful when we talk to patients about treatment based on K-ras testing, because basically we’re taking away treatment. The good part is that they may not need to be exposed to side effects of a therapy that will not help them, but their initial reaction is to retract.
We dealt with this at ASCO when we had an emergency meeting about the Intergroup trials and our patient advocates were there. Their response was that we had taken too long to do this and we explained that we didn’t want to change all the trials based on early verbal reports. We needed them to understand and buy into the fact that we’re not denying anyone anything. It’s like denying paclitaxel to a patient with colon cancer — it doesn’t work. If we could identify a subgroup who did respond to paclitaxel or gemcitabine, that would be nice. But it would be a small group, because it never shows up in any analysis, not even a smell of something. Patients want to be active. They want to be proactive and go in there and receive it.
