Here is the second case in my expert round table discussion on locally advanced NSCLC with medical oncologist Dr. George Blumenschein froMD Anderson Cancer Center and radiation oncologist Dr. Walter Curran from Emory University.

We focus in this case on the decision of which patients with a Pancoast tumor should undergo surgery or a nonsurgical approach of chemo/radiation, the challenge of trying to define the right time to repeat scans after chemo/radiation in locally advanced NSCLC, and we also debate the merits of close observation vs. further interventions in the face of worrisome but still ambiguous imaging findings.

I’ll add that I do find it instructive how varied the advocated treatment approaches are among the various experts when discussing not only this case but so many others I present.  These are admittedly challenging cases that don’t fit into any “classic” treatment approach, but these discussions of the range of alternatives offered by experts from so many places speaks to the fact that there is rarely one best strategy.

As always, here is the podcast in both audio and video formats, along with the transcript and figures.

imaging-issues-after-chemoradiation-case-blumenschein-curran-audio-podcast

imaging-issues-after-chemoradiation-case-blumenschein-curran-transcript

imaging-issues-after-chemoradiation-case-blumenschein-curran-figures

Read the rest of this entry »

Podcast: Play in new window | Download

A radiologist, the person who specializes in reviewing imaging studies in medicine, is often someone you notice if they’re unusually bad or unusually good.  They perform a service and you presume that they’re good at it, but a few are so sharp that the other doctors they work with notice it at every tumor board discussion or one on one exchange.

Dr. Manning is one of those special radiologists.  She had trained at the University of Washington while I was completing my medical oncology training there, and when she left to join Seattle Radiology, a private group based at Swedish Hospital in Seattle, I was among the many doctors who felt the body blow of losing her talent.  On the other hand, when I later joined the medical oncology group at Swedish, I was very happy to have the opportunity to be reunited with her.

But talk is cheap, so I’ll also add that when I developed a cough that lasted for months a few years ago, and one of the thoracic surgeons I work with said I needed to get a chest CT to check it out (as I joked that his practice was slow and that he was trying to drum up business), it was Dr. Manning who read my own scans for me (and fortunately, they were OK).

She remains a terrific resource, and one too valuable to not share with others.  She was kind enough to sit down with me for a discussion of current issues in imaging, with a particular focus on issues related to lung cancer.  Here’s the first part of our discussion, which covers a bit on screening, the issues related to assessing and following lung nodules, and some basics of the work-up and ongoing follow-up of patients with lung cancer.   Below you’ll find the audio and video versions of the podcast, the transcript, and a copy of the very few figures associated with the audio component:

dr-manning-imaging-intro-part-1-audio-podcast

dr-manning-imaging-intro-part-1-transcript

dr-manning-imaging-intro-part-1-figures

Read the rest of this entry »

Podcast: Play in new window | Download

   As I described in a post last year, one of the common features of angiogensis inhibitors is that lesion often cavitate, shrinking not only from the outside in, but hollowing out and dying from the inside out.  One of the concerns has been that cavitating lesions may be at higher risk for bleeding, but the work I described in the prior post suggested no clearly increased risk of severe bleeding complications among patients who develop tumor cavitation on anti-angiogenic agents.  But another issue is that tumor cavitation may be a relevant measure of response that is being missed by out current systems of measuring response by only assessing the outer dimensions of a tumor.

    Investigators in Ontario, Canada looked at this issue by assessing the frequency of this cavitary response in a couple of recent trials they ran through the National Cancer Institute of Canada (abstract here).  This report looked retrospectively at 33 patients who received doublet chemo with an antiangiogenic agent (in this case, the oral VEGF inhibitor cedirinib, also known as AZD-2171) and found that 8 (24%) experienced tumor cavitation during treatment.  This wasn’t seen in any of the 18 patients treated with chemo alone on another trial of chemo alone that ran over a similar time line in Canada (and it’s rarely seen with chemo alone in general).  

   A typical situation is shown in the figure below, in which a mass doesn’t change outer dimensions, but it hollows out early in treatment, then progresses by the filling in of the center of the tumor. 

It’s probably very relevant that the overall volume of viable cancer has decreased substantially early in treatment, but that response isn’t captured by our current system.  Nor is the clear increase in tumor bulk as the cancer grows internally, since outer dimensions again don’t change.

   The investigators tested how many patients would have experienced a change in their official response if an alternative definition of response was used in which tumor measurements included not just outer dimensions but subtracted the hollowed out portion of any cavitating lesion.   And progression would be defined not only by outer dimensions but by the filling in of cavitation:

   When they did tumor measurements by both the current standard (outer dimensions only) and the new alternative that incorporates measurement of the tumor cavity, a few of the patients had differences in their response assessments, going from stable disease to a partial response, or from stable disease to progression.  They started with small numbers of patients and just assessed these issues retrospectively, so you really couldn’t say anything definitive.  But it was interesting to see that the current system probably misses some meaningful anti-angiogenesis-induced responses and can keep people on an ineffective therapy for too long while they progress within the stable shell of a tumor.

   The invstigators suggested that this new technique needs to be tested in trials moving forward, and I agree. But even this early work could be useful

      Continuing on the introduction to the concept from a recent prior post, the issue of whether it’s important to see an improvement in progression-free survival (PFS) if there is no improvement in overall survival (OS) after additional therapy is going to be a central issue in lung cancer management, relevant in several key issues in coming years.  This is actually a fortunate development, because it’s a by-product of there being enough valuable treatments to give sequentially that we’re questioning whether the order matters at the end of all of the treatments a patient will ultimately receive.  The breast cancer community has been grappling with this issue for years, since there are a wide range of effective treatments for advanced breast cancer.  Though two drug combinations have a higher response rate and PFS, that doesn’t translate to a higher OS when breast cancer patients typically receive more than four lines of treatment over time, so a typical approach is with sequential single agents rather than an urgent need to treat with the most aggressive therapy first.  They also typically start with hormone therapies, which are often somewhat less active in terms of response rates and PFS but have an advantage of being associated with fewer side effects compared to standard chemotherapy.

    Ten years ago it was actively debated whether there was a value in treating metastatic lung cancer, given how marginally effective and more challenging the treatments were.  The concept of second line therapy really only took off around 2000-2001, when taxotere (docetaxel) was shown to improve survival, again modestly, and a challenging treatment as well.  It’s really only been in the last five years, with the addition of alimta (pemetrexed) and the EGFR inhibitors iressa (gefitinib) and then tarceva (erlotinib) were introduced as additional options with activity in previously treated patients, and a side effect profile that patients and oncologists could more readily accept, that further lines of therapy became standard and anticipated.  While 6-7 years ago there were almost no trials of new drugs being tested as second line treatments (alone or added to taxotere, for instance), there are now dozens of large trials of new agents being tested as second line, third line, even some fourth line options.

   So we now need to struggle with the optimal order of the growing number of tools available.  Is it better to use them combined early, or as single agents sequentially?  And does the order matter?  This issue is now at the center of our current debates. 

Read the rest of this entry »

   One of the topics that frequently occurs in the clinic, and that patients often ask about, is the situation in which there is some suggestion of slight progression.  This can take the form of many different situations:  a rising tumor marker (see prior post), a slight increase in the uptake on PET (see prior post), perhaps an increase in cough or pain between CT scans, or perhaps just a scant amount of disease progression on the CT that isn’t enough to lead to a definite transition to some alternative.   This isn’t an area where there are any good studies, so we are left with our best judgment and no substantial information to inform us.  

Read the rest of this entry »

   Intuitively, you’d think that people who are doing worse while getting treated for lung cancer are not going to do as well as people who have improvement in their symptoms after treatment starts.  But how much do patient symptoms count in our current medical system for deciding whether a treatment is working or not, and when to move to a new therapy?  The answer is that patient reported symptoms don’t have a clear role yet.  That has partly been because developing and using validated, predictive symptom scales can be expensive and challenging to get full participation, these scales haven’t been clearly correlated with more established efficacy measures like response rate on CT scans and overal survival, and it hasn’t been clear whether the FDA would accept such measures in appoving a new drug.  A drug that is shown to improve survival is highly likely to be approved, since most doctors and many patients consider how much a treatment improves survival to be the most important benefit a cancer treatment can deliver.   And unlike many symptoms that can be subject to a placebo effect and can only be assessed by the patient, survival is a “hard endpoint” that isn’t really subject to any debate.   But obviously survival is only one aspect of the treatment experience, and it’s not something you can get early feedback on. 

   But there’s a growing momentum for incorporating patient feedback as an early indicator of the value of a treatment.  I’ll write a few posts in the coming days and weeks about the challenges nd developments of patient-reported symptoms and quality of life assessment in oncology.  In the meantime, I’ll start with a report that demonstrates that patient symptoms are correlated with our conventional measures of efficacy.

   Filippo de Marinis and colleagues did a retrospective review (abstract here) of an important phase III trial in advanced NSCLC that I’ve described in a prior post, directly comparing taxotere to alimta as a second line treatment.   To summarize that trial in one sentence, it showed that alimta and taxotere had essentially identical activity, as well as quality of life, but a few of the blood-related side effects were significantly less with alimta; this led to the FDA approval for alimta as a second-line treatment for advanced NSCLC.   Of note, in this trial, as well as all of the other large trials of second or third line treatment, the response rates are very low, in the 6-10% range, despite the apparent benefits in terms of improved survival from these agents for a broader population.  Read the rest of this entry »

   A cancer has to grow faster than the tissue around it to become a tumor.  Progressive growth is therefore a central feature of a cancer and a critical factor in distinguishing cancerous nodules from benign ones.  There is a characteristic “volume doubling time” (VDT), the interval it takes for a nodule to double in volume.  It’s worth keeping in mind that because a nodule is generally spherical, an increase in the diameter by just 28% (such as a 2 mm increase from 7 to 9 mm) actually represents a doubling of the volume of a nodule.  

   So we expect a cancer to grow, but there’s a lot of variability in the rate of an individual cancer’s growth.  We know that different histologies (cancer subtype under a microscope) have different growth characteristics: SCLC has a typical VDT of 30 days, while NSCLC has an average VDT of 100 days.  But on one end of the spectrum for NSCLC, a poorly differentiated adenocarcinoma or squamous cell carcinoma may have a VDT in the range of SCLC, while a well-differentiated cancer like many bronchioloalveolar carcinoma (BAC) tumors may have a VDT of more over a year.   So by observing a nodule over time, we can learn a good bit about the probability of it being a cancer.   For instance, nodules that grow incredibly quickly (VDT less than a week, for instance) are far more likely to be infection than cancer, and those that go for a year and a half or so without doubling are generally too slow-moving to be cancer.  The common definition of stability is a lack of change over two years of observation, and then patients with a nodule being followed are generally felt to have a very low likelihood of cancer.   However, I’ve seen a few cancers that didn’t have convincing growth until more than two years of follow-up had been completed (although a cancer that takes 3-4 years to double is likely to be much less life-threatening over the next several years than a more typical, faster growing lung cancer).  And there’s always a rare very rapidly growing cancer that can grow remarkably during a period of observation.  The key is weighing the risk of watching (which usually helps to clarify the risk that a nodule really is cancer or not) vs. the risk of jumping in and overtreating lots of nodules that aren’t cancer after all.

   Among the key principles is that it’s not sufficient to just “eyeball” a nodule, but rather to have a radiologist carefully measure the nodule on a scan.  And the level of detail required necessitates follow-up CT scanning rather than chest x-rays.  Even with measuring nodule diameter on CT scans, very small nodules of just a few mm can differ in their apparent size just as a function of where the CT slices through the nodule (slice cuts may be every 3-5 mm for example).  Imagine the difference between cutting through an orange through the middle vs. toward one side.  If you only see the cross-sectional size, different cuts can give a different sense of the same nodules.

    As I suggested in the discussion above about how tumor histology correlates with VDT, there have been studies that have looked at different kinds of tumors and their typical doubling times.  For instance, ground glass opacities typically have the slowest VDT, with solid nodules a faster VDT, and ”semi-solid nodules” in between the two (abstract here).   In this study, the median VDT for pure GGOs was 813 days (as in, more than two full years), underscoring the point that some of the most indolent cancers, predominantly BACs, can grow remarkably slowly but still be a cancer (see prior post here about some of the issues with the most indolent BAC tumors). 

   Finally, there’s a website that includes a doubling time calculator, in which you input the dates of your CT scan studies and the size of the nodule.  The calculator is here (and there’s also some more information on imaging of lung cancer).   

   The key points, though, are that cancers should grow over time of being observed on CT, and while there is a lot of variability in how quickly lung cancers grow, the fastest and slowest moving are less likely to be cancer than the ones that double over something in the range of a month to a year or so.  The ones with convincing but not extremely rapid growth are the most suspicious for cancer.

  As you might suspect, there are features of different spolitary pulmonary nodules (SPNs) that makes us more or less suspicious for cancer.  The first is the size of the nodule.  Looking at multiple series of SPNs, the likelihood of cancer among nodules that measured under 5 mm is generally in the 0-1% range.  Nodules in the 5-10 mm range have been found to be cancer in up to about about 28% of cases, with most studies showing the risk of cancer in this range to be one in four or five.  SPNs in the 1-2 cm range are found to be cancer about 40-50% of the time, and beyond that, the risk is up to about 60-80%.  But even among the larger ones, that go from being a nodule to a mass (3 cm cutoff), the likelihood of cancer isn’t 100%, and there are many things that turn out to be infection or inflammation or another benign cause.

   In general, smooth borders for a nodule are somewhat reassuring, especially solitary ones.  Metastases from other places tend to be smooth and toward the outside of the lung, so that’s an exception of when smooth edges can be a bad thing.  The most common shape of a primary lung cancer is spiculated, or stellate, with little projections, so it has irregular borders.

   Then there’s the issue of the density of the nodule.  Solid nodules appear as pretty uniformly white on the CT (example in my prior post), while those that are just hazy clouds are called ground-glass opacities, or GGOs. In between is a nodule that appears  to have areas of non-solid haziness next to other areas that are solid.  These are called, coincidently, semi-solid nodules or opacities.

(Click to enlarge image — but it still may be hard to see the haze around the nodule on the right image) Read the rest of this entry »

   We know PET scans can provide additional metabolic information that can be more sensitive and specific for cancer than chest x-rays and even CT scans in the initial staging of lung cancer (see prior post on introduction to PET scans). PET scans are now nearly universally employed in the initial workup, at least of patients who have NSCLC and aren’t already known to have stage IV disease. But how useful is this technology in the setting of surveillance for the patient at risk for recurrent/residual disease after curative treatment?

   Unlike CT scans, which are great at discerning shape and size of internal parts of the body, PET scans are metabolic studies that hold the promise of distinguishing between residual viable cancer and non-viable scar tissue after surgery or radiation. A handful of studies in the post-treatment setting show that PET scans have a 96% sensitivity (PET scans identify a problem 96% of the times that there is a problem) and an 84% specificity (there is viable cancer 84% of the time that the PET scan indicates that it’s there) (example references here and here and here). The problem, though, is that unlike the situation of initial staging, where patients haven’t undergone interventions that could cloud the interpretation of a PET scan, except perhaps for a recent biopsy), after surgery or radiation, the inflammatory changes after treatment can also light up on PET scan and make it difficult to interpret what’s cancer and what’s post-treatment inflammation or infection, especially in the vicinity of the original treated cancer. Because of this concern, which is greatest shortly after local treatment, people have generally recommended that PET scans not be pursued until at least 3-6 months after treatment because of the high risk for false positive results (scan shows abnormalities that aren’t really cancer), and also that any areas that appear suspicious on a PET should be confirmed as cancer with CT imaging and a biopsy before pursuing any interventions against presumed cancer.  The concept that PET scans will improve our ability to detect curable recurrences or new cancers hasn’t yet been supported by any evidence of better survival or quality of life in patients.

   In the face of such a confusing situation, there’s a lot of variability in what oncologists actually do. As a rule, I don’t obtain regular PET scans on patients after curative surgery or chemoradiation, and I’m particularly reluctant to do so in the first few months after completion of treatment. For my patients with a better prognosis, such as stage I and II NSCLC, I routinely obtain CT scans and will order a PET scan if the CT shows something suspicious (typically I’ll get a PET/CT fusion scan 4-6 weeks after the concerning CT to see if the suspicious area has changed and whether it appears as metabolically active on PET). For stage III patients, in whom the risk of recurrence is higher, I do tend to obtain a PET/CT about once per year, otherwise following with CT scans. My general philosophy is that PET scans can be so sensitive but non-specific, especially after chemoradiation with or without surgery for stage III NSCLC, that changes in the absence of CT findings can too often be an anxiety-laden wild goose chase. Chest CT scans provide a lot of detail, so if there isn’t anything suspicious that has emerged from one CT to the next, I’m quite reassured that anything you might find on a PET scan would have a high chance of being a false positive.

   This is an area, though, where we’re gaining more experience, and I really hope and expect that we’ll actually get some study results to help guide or practice patterns. In the meantime, it’s the open frontier (the Wild West if you prefer) in terms of using PET scans for surveillance, with people just following their own rules.

   Among the key issues in following patients with a history of treated lung cancer is the pattern of recurrence. We need to have a sense of when the risk is highest and where people are more likely to demonstrate new evidence of disease. Fortunately, there are several studies that can help us with these questions.

   Looking at patients who have undergone potentially curative surgery for stage I NSCLC, we know that recurrence rates in the range of 20-35%. Most of these recurrences are distant metastatic disease, not failure of local treatment that leads to disease recurrence close to the site of the original primary. This is why we feel that chemotherapy, which goes throughout the body, is the more effective way to reduce the risk of cancer recurrence after surgery.  In addition, most recurrences are in the first 3-4 years after surgery, but there are also reported recurrences beyond 5 years out. We know that people who have nodal involvement have a higher risk of recurrence than those with stage I disease (and no lymph node disease), and they also tend to develop recurrences earlier than those without nodal involvement.

   In addition to the risk of recurrence of a prior cancer, there is also a risk of a new cancer developing. We know that patients with a history of lung cancer, not only have a risk of their prior cancer returning, but they also generally have a greater predisposition to develop another, independent lung cancer.  In many cases, the cells of the lung have a field defect (a problem that affects not just the cancer, but an area around the cancer as well), often from prolonged exposure to carcinogenic tobacco smoke, which means a large population of cells is also part way along a path toward cancer. The estimate of risk for developing a new lung cancer is in the range of 1-2% per year, and this remains constant with more time after potentially curative surgery for a first cancer. Read the rest of this entry »