Last night I had the good fortune to attend the fun young lung dinner. I had a lot of fun seeing old friends, and greatly enjoyed making a few new ones:

Now, after two double espressos, I’m excited to be sitting in the main plenary session, which focuses on the very best talks of the whole conference.

Squamous Cell Lung Cancer: Genomics and Targeted Therapy by Peter Hammerman Md, PhD of DFCI

Dr. Hammerman started by reminding us of the nature of cancer; specific types of genome alterations (mutations/changes in DNA) drive cancers. We’ve talked about this a lot in adenocarcinoma, where we can identify a driving oncogene in about half of cases. However, SqCC is also driven by mutations and this will be the focus of Dr. Hammerman’s talk.

Until very recently, our knowledge of driving mutations in SqCC was very limited. Only PIK3CA and BRAF have strong data thus far for driving SqCC, their presence is rare, and we have no proven effective therapeutics against either in lung cancer. For these reasons, the cancer genome atlas project has focused on SqCC. Each sample received undergoes a very rigorous quality control procedure, then complete molecular characterization including DNA sequencing of all exons, RNA sequencing, array-based assessments, and epigenetic analysis.

There are now a number of gene copy numbers that are altered in SqCC:

Gene	Event Type	Drug target
FGFR1	Amplification	Yes
SOX2	Amplification	No
CCND1	Amplification	Yes
REL	Amplification	No
PDGFRA	Amplification	Yes
EGFR	Amplification	Yes
NFE2L2	Amplification	No
MCL1	Amplification	Yes
ERBB2	Amplification	Yes
CDKN2A	Deletion	Yes
PTEN	Deletion	Yes
RB	Deletion	No

And there are mutations:

Gene	Number of patients	Therapeutic Target
TP53	104	No
CDKN2A	23	Yes
KEAP1	19	No
LPHN3	22	No
NFE2L2	16	No
RB1	15	No
PIK3CA	13	Yes
PTEN	11	Yes
NOTCH1	10	Yes
ERBB4	10	No
NTRK3	8	Yes
FBXW7	7	No
FGFR2	6	Yes
DDR2	5	Yes
PDGFRA	5	Yes
BRAF	4	Yes
HRAS	3	No

As cancers go, lung cancers are particularly genomically complex—they are second only to melanoma. In contrast, more curable cancers like hematologic malignancies of childhood seem to be less complex. In looking at the table of different cancers and their levels of complexity, a clear theme emerged that cancers driven by known carcinogens like head and neck cancer and lung cancer tend to be more complex.

Now, we can identify a possible therapeutic target in 63% of SqCC. In particular, FGFR1/2, PIK3CA and DDR2 inhibitor trials are moving towards the clinic for trials.

Gene	Event Type	Frequency
FGFR1	Amplification	20-25%
FGFR2	Mutation	5%
PIK3CA	Mutation	9%
PTEN	Mutation/Deletion	18%
CCND1	Amplification	8%
CDKN2A	Deletion/Mutation	45%
PDGFRA	Amplification/Mutation	9%
EGFR	Amplification	10%
MCL1	Amplification	10%
BRAF	Mutation	3%
DDR2	Mutation	4%
ERBB2	Amplification	2%

Dr. Fong discussed the abstract. He attributed success to improved scientific methods, increased funding from the stimulus act, and decreased cost of sequencing. He believes that we will need a new language to discuss the many molecular changes that happen in complex cancers such as SCLC and SqCC. He expressed optimism that we will ultimately improve therapy with this understanding, but warned that the path will be hard. There are many different mutations in complex cancers and not all of them are actually driving the cancer. Not every SqCC is driven in the same way. We will need understand resistance mechanisms. We will ultimately want to personalize therapy by both the tumor genome (to predict which drug will work best) and host genome (to predict toxicity).

Improvements in survival of elderly patients with stage I NSCLC in the Netherlands between 2003 and 2009 by Dr. Cornelis Haasbeek

In recent years, there have been advances in the treatment of stage I NSCLC in patients unfit for surgery. Stereotactic ablative radiation therapy was introduced in the Netherlands in 2003 and local control rates have been higher than 90%.

Dr. Haasbeek and colleagues evaluated the Netherlands cancer registry which includes individual data on all Dutch Cancer patients. They evaluated three time periods: 2001-2003 (before intro of SABR), 2004-2006 (limited SABR availability) and 2007-2009 (full SABR availability). All patients had stage I disease and were at least 75 years old. There was no shift in the use of surgery. However, there was a shift from no treatment to SABR.

Time Period	Surgery	Radiation	No Curative Treatment
2001-2003	37%	31%	32%
2004-2006	36%	33%	31%
2007-2009	37%	38%	25%

Overall, the median survival improved from 16.4 months to 24.4 months. In the surgical group, median survival improved from 35.7 months to not yet reached. In the radiation group, median overall survival improved from 16.8 months to 26.1 months.

The improvement in the surgical group was possibly attributed to improved perioperative care, decreased operation on high-risk patients (switch to SABR), and improved surgical technique (including VATS). Similarly, the radiation group may have done better secondary to the introduction of SABR (in place of older radiation techniques).

Dr. Hak Choy of UT Southwestern discussed the abstract. He was very funny, noting that surgery is much “sexier” and faster than radiation and has more intuitive appeal to some patients. He reminded us of the history of radiation therapy including that fractionating radiation (delivering it over longer times instead of all at once) has decreased toxicity. However, recent advances allow greater targeting of the radiation, allowing us to come back towards faster (hypofractionated) radiation.

Dr. Choy noted that there is literature to back up these findings. In particular, he reminded us of the excellent results in the RTOG 0236 study.

In vivo selection of NSCLC patients with activating mutations in the tumor EGFR using [11C] erlotinib and PET by Dr. Idris Habce of the Netherlands.

Dr. Habce started by reviewing data on EGFR and the efficacy of TKIs against it. He went on to note how hard it is to get good tissue in lung cancer. The basic idea was to radiolabel erlotinib, give it to patients, then see where the tracer went through PET imaging. The hypothesis was that because erlotinib has a greater affinity for mutated EGFR than wild type EGFR, mutant tumors would light up brighter.

5 patients with wild type EGFR and 5 patients with exon 19 deletion were imaged. There was a much higher uptake for patients with mutation, with p.03 and little overlap between the two groups. The results held for treatment results: 2/5 patients with wild type (non mutated) tumors progressed. 3/4 patient with mutation achieved PR. 1/5 patients with PD had mutation; interestingly, this patient had the lowest uptake of the mutant patients.

Dr. Paz-Ares from Sevilla, Spain discussed the abstract. He noted how hard it is to get tissue in lung cancer and noted the lack of validation of proposed alternative methods for predicting who will benefit, including serum markers and CTCs. He reminded us that C-MET PET has already been studied in recurrent glioblastoma to guide therapy and trastuzumab PET has been used in breast cancer. He reviewed what makes a good PET probe and judged that this one meets the bar. He also reviewed other tracers under development, including cetuximab and irreversible TKIs. He pointed out that regulatory approval of these kinds of tests will be challenging and expensive. Finally, he made the excellent point that we need to test for multiple molecular markers, which, at the current time, requires tissue.

The effects of investing in thoracic surgery on lung cancer resection rates by Dr. Lau of the UK, by Dr. Kelvin Karin Wing Lau of the UK

Lung cancer in the UK is undertreated and routinely has some of the poorest outcome results in the Western world. Resection rate varies dramatically within England. Dr. Lau’s group hypothesized that the variability in resection rate is determined by the provision of specialists in thoracic surgery (as opposed to surgeons that do both lung cancer and cardiac surgery or even general surgeons). 58% of hospital trusts had less than 2 pure thoracic surgeons. Resection rates were very low, even in early stage disease. Resection rates tended to be higher in centers that treat more cases. When patients first presented at a major center, their resection rate was higher; when referral was required, resection was lower. Resection rates are higher in centers with two or more specialized thoracic surgeons. In 2009, 5 of 31 units expanded their number of specialist thoracic surgeons. These centers dramatically increased their resection rates. Put together, this data suggests that specialized thoracic surgical availability drives greater rates of surgery in England.

Dr. Lau showed unpublished data from the National Cancer Intelligence Network showing higher survival in centers with greater volume:

Resection Quintile	Adjusted HR (survival)	95% Confidence Interval
1 (Highest)	0.88	.86-.91
2	0.93	.9-.94
3	.93	.91-.95
4	.98	.92-.97
5 (Lowest)	1

P<0.001

Methods in Mutational Analysis

Next, I went from nerdy to super nerdy. If you don’t do mutational analysis properly, you don’t get the right results, and therefor decrease your chance of making the right treatment decisions. As a medical oncologist, what happens in the lab is often a bit of a black box. I have the good fortune to be friends with a molecular pathologist, who has opened up that black box a bit for me, and I decided to attend this session to try to better understand the issues.

EGFR testing in Lunc Cancer: experience and suggested guidelines; techniques and quality control by Dr. Marc Ladanyi of MSKCC, NYC, USA.

Classically, EGFR and KRAS mutations are not seen in SqCC, classic SCLC or neuroendocrine LCLC. However, as I type this, I am sitting next to Dr. Dara Aisner of the University of Colorado. This afternoon, she will present two cases of proven SqCC (on surgical resection) which also have EGFR mutation. Her group has data showing a 4-5% rate of EGFR mutation in SqCC. She also noted that with small biopsies, components of adenocarcinoma can be missed, increasingly the rationale for molecular testing in patients with clinical characteristics that predict the presence of mutation.

Dr. Ladanyi reviewed data on smoking and EGFR mutations to show that former smokers with adenocarcinoma do have mutations (13.5%), and even current smokers can (4.9%).

Dr. Pao and colleagues, in a Clinical Cancer Reviews paper in 2007 showed how different mutation detection methods have different operating characteristics:

Unmodified Sanger sequencing (the traditional way of detecting mutations) will not reliably detect mutations if the proportion of tumor cells is <25%. When you consider that most of a biopsy consists of normal tissue, with tumor making up a small proportion, you realize how easy it is to get below 25%. Formalin-fixed paraffin embedded tissue is better than frozen tissue for assessing tumor content.

How can we improve this? One approach is micro dissection—dissecting tumor away from normal tissue under the microscope. When the tumor is well separated from the normal tissue, this technique can be powerful. However, if they are intermixed, it can be challenging. It is also operator dependent—this is something that a superb molecular pathologist or technologist can do much better than someone with less experience.

Another approach for improving results, including in tumors for which micro-dissection is not helpful, is a technique called locked nucleic acid PCR. Locked nucleic acid is a modified nucleic acid with increased affinity to complimentary DNA. It can be used to “block” the wt (normal) DNA, so that when you run mutation testing, you will be more likely to find a mutation, if present. He mentioned that there is another technique, called “cold PCR” (coamplification at lower denaturation temperature for fellow nerds out there). I realize that I’m not fully explaining the details of these complex techniques—for now, I think that it’s good enough to say that they can improve sensitivity, or the chance of finding a mutation if it really is there.

Dr. Ladanyi reviewed quality control measures, which I will not review here other than to say that they are important.

Detection in Blood/Serum by Dr. M. Taron

As getting biopsies can be hard in lung cancer, we’ve talked before about blood-based testing, including circulating tumor cells and circulating DNA. So, clinically, it is of great interest to spare patients biopsy and to get molecular results in patients for whom biopsy is impossible. Again, however, technique is an issue. A blood-based technique called cfDNA has sensitivity of about 50%, meaning that it will find 50% of mutations that are present. In patients treated with a TKI, real clinical data showed better PFS and OS with mutation detected by this method; the data looked very similar to what we’ve seen with mutation testing by traditional sequencing. To my mind, a test like this is more helpful when positive than when negative. If it’s positive, it could lead to treatment with erlotinib or gefitinib. When negative, the molecular status must still be considered unknown. Dr. Taron also noted that blood based testing could allow for safe serial assessment of molecular status. For example, he showed a patient whose blood levels of the T790M resistance mutation increased prior to progression, with levels declining after successful chemotherapy.

Heterogeneity in EGFR in Lung Cancer by Dr. E Thunnissen

GRACERs, please forgive me my lapse of attention during much of this presentation, as I’m sure that it was great—I’m fairly caffeine deficient at this point but did get the take home point and will share it with you. Several papers have shown that primary tumors and metastases can have different results on mutational analysis. This presentation argued that many of these discordant results may be due to limitations in methods, as opposed to genuine differences between primary and metastasis. In the question/answer session, another pathologist in the audience pointed out that another potential explanation is multiple primary tumors.

Bimodality and Trimodality Therapy

As we’ve discussed here before, the treatment of stage III lung cancer is one of the most controversial areas of lung cancer. A variety of treatment options are reasonable and many experts feel passionately that their preferred method is the best one. My opinion is that the available data do not define one best standard of care and so I was anxious to go to this session and see if any new data would influence my thinking.

Pre-operative chemotherapy improves survival and reduces recurrence in operable non-small cell lung cancer: preliminary results of a systematic review and meta-analysis of individual patient data from 13 randomised trials by Dr. S. Burdett

This trial is a meta-analysis of 13 trials including 2094/2581, or 81% of all known randomized pts. Overall, treatment was fairly heterogeneous—these trials gave very different treatments from each other. Overall, 5-year survival was improved from 30% to 40% (HR .88, p.025). Both local recurrence and distant recurrence rates were reduced. These numbers are similar to the numbers we’ve seen for postoperative therapy, enforcing my opinion that they’re probably about equivalent.

RTOG 0229:A Phase II Trial of Concurrent Chemotherapy and Full Dose Radiotherapy Followed by Surgical Resection and Consolidative Therapy For Locally Advanced Non Small Cell Carcinoma of the Lung by Dr. M.J. Edelman

This trial evaluated full dose radiation with chemotherapy, as opposed to previous trials that stopped at lower radiation doses (61.2gy instead of 45gy). The chemotherapy was weekly carboplatin and paclitaxel during radiation, with full dose chemotherapy given following surgery. PFS at 24 months was 33%, median OS was 26.6, and 63.3% of patients had mediastinal clearance (all cancer in lymph nodes in the middle of the chest killed with chemorads) . As we’ve seen before, results were better for patients who had mediastinal clearance than those who did not. The trial shows that full dose radiation is feasible when combined with chemotherapy and surgery.

Competing risks analysis and long-term survival (LTS) of patients (pts) with locally advanced (LAD) stage III non-small-cell lung cancer (NSCLC) treated on sequential trimodality (TM) trials by Dr. W. Eberhardt

The authors looked back on four trials conducted between 1991 and 2002 to identify subgroups that seemed to benefit of be harmed in order to improve future trials. There was no major difference between patients with Stage IIIa and IIIb disease, although there was a trend towards lightly better results with IIIa. Patients younger than 50 had a trend towards better results. Patients with SqCC had much better results than patients with adenocarcinoma, with a p<.0007. Overall survival was better with carbo/taxol chemo than with cispaltin/etoposide chemo. Patients with R0 (complete resection) did better than patients with R1 resection (microscopic disease left behind). The most remarkable results were the comparison of SqCC vs. adenocarcinoma and I look forward to seeing if these results can be confirmed elsewhere.

Is there an Optimal Concurrent chemo-RT regimen for Stage III NSCLC? -- Preliminary analysis of ACRIN 6668/RTOG 0235 by Dr. M. Machtay

Over the years, RTOG trials have used different chemotherapy regimens in chemoradiation. RTOG 0235 aimed to look at SUV on PET after treatment as a biomarker. 250 patients were evaluated with PET both before and after chemoradiation. Patients and their doctors were allowed to choose the chemotherapy used, and this analysis compared the two most common regimens: carboplatin plus paclitaxel and cisplatin plus etoposide. 93 patients were treated with carboplatin plus paclitaxel and 35 patients received cisplatin plus etoposide. There were no significant differences in survival between the two regimens. When radiation dose was evaluated, higher dose did seem to help, at least up to 70Gy.

Results of a phase II trial on individualized radiation dose-escalation based on normal tissue constraints in concurrent chemo-radiation for stage III non-small cell lung cancer (NSCLC) (NCT00572325 trial) by Dr. A. van Baardwijk1

In chemoradiation, there is debate about optimal radiation dose. As you increase dose, tumor control increase, but so does toxicity. Most trials compare a lower dose to a higher dose, and the results are mixed. This trial personalized radiation dose by evaluation of each patient’s tumor and normal anatomy, giving the maximal dose felt safe for the specific configuration of each patient’s tumor relative to important normal structures. Median dose was 74Gy. Median overall survival was 24.2 months, with 2 year OS of 52%. 2 year PFS was 40.3%, with a median of 16.5 months. Toxicity was acceptable.

Overall, the combined modalities session didn’t really change my feelings on the subject. I still think that there are many reasonable options. I still feel that, with the current state of the data, the only major mistake that can be made is to think that we know what we don’t.