Our CME group constantly seeks new methods to delve into the deepest reaches of the minds of clinicians to tease out precisely how these individuals care for patients with cancer. Our most recent experiment in evidence-based psychology is a little tune we call “second opinion.”

In many surveys and interactive polling conferences, we have queried docs about their usual recommendations for patients in various clinical situations. For this issue of Patterns of Care, we add a new twist to this time-tested strategy and focus on how physicians react to the recommendations of other docs.

The purpose of this exercise was to identify situations where concordance and discordance exist in the application of evolving clinical research findings.

For example, for the enclosed October 2007 survey, we presented the cases of three women with node-negative breast cancer to our cohort of 150 medical oncologists and 50 research leaders, and specified a “first opinion” from a different source for each case.

We then asked these individuals if they agreed with the first recommendation and, if not, whether they felt strongly enough to state that the first opinion was not an acceptable option.

It is an interesting reflection of how this field has developed that the approaches to these three patients vary based on variables like HER and ER status and patient age.

The responses suggest some strong disagreement in clinical practice that have profound implications for patient care. In essence, we found situations in which the preferred treatment recommendations of a substantial number of docs are not acceptable to many others, and these relate to major decisions such as the decision to recommend adjuvant chemotherapy or not.

Below, find a few thoughts on these test balloons, including a “third opinion” that reflects my views of how the risk-benefit tradeoffs might look to me as a patient.

Way back in 2000, the NIH Consensus Conference on Adjuvant Therapy for Breast Cancer would not have recommended adjuvant chemotherapy for this woman, based on the small size of the tumor. What the consensus panel had no way of predicting was that after decades of less-than-exciting studies and papers on prognostic and predictive factors in breast cancer, someone finally got it right in the form of the Oncotype DX assay.

How much has this test changed practice? For this case, we noted that the recurrence score was high but that the first doc recommended only endocrine therapy — a standard approach in 2000, although in this case, the treatment recommended (an LHRH agonist and an AI) would not have been considered at the time.

Today, almost all investigators and practicing oncologists would use chemotherapy in this situation, and most feel so strongly about it that they essentially would refute the first opinion for withholding chemo.

The choice of hormone therapy in this case is also interesting in that for this woman, who was premenopausal at diagnosis with a node-negative tumor, most docs would use an AI at some point, either up front with ovarian suppression or after menopause had interceded.

One important aspect of the NSABP data set on Oncotype DX is that patients with high recurrence scores generally derived much less or no benefit from tamoxifen. As such, clinicians may be taking this as a clue to consider new or more promising therapies — for example, an AI and LHRH agonist — even without definitive trial data.

In 2008, the Oncotype DX assay will, for the first time, include reporting of quantitative ER and PR, and it will be interesting to observe whether docs start altering decisions in challenging cases like this one based on these numbers.

As the third opinion on this vexing case, here’s what I might be thinking if I were the patient looking at a prognosis without chemo similar to a patient with a node-positive tumor and with the potential to reduce my risk of recurrence by 75 percent. I’d force myself to consider chemotherapy, and like the majority of survey respondents, I’d go with TC (docetaxel/cyclophosphamide) — another intervention that pretty much didn’t exist in 2000.

In terms of hormone therapy, I’d like to see that quantitative ER, but even if it was high, I would likely be nervous that this was an aggressive tumor that just got caught early. If I were not interested in childbearing — and at 45 and about to receive chemo followed by hormones for at least five years, that might be a far away thought — perhaps I’d just check into the closest laparoscopy center for an oophorectomy and maybe add tamoxifen or more likely an AI and see how it goes.

These are the same numbers as in case 1 but for a postmenopausal woman. The difference here is that the first opinion recommended chemotherapy (dose-dense AC) but the endocrine treatment suggested was tamoxifen.

As with the first case, most docs didn’t much like the first opinion, but in this case, mainly due to the endocrine recommendation. It’s interesting to reflect back to December 2001, when Mike Baum presented the first AI adjuvant data (ATAC) in San Antonio. On that day, Mike and Aman Buzdar, another ATAC trialist, in separate interviews both flat out said, “It’s time to say goodbye to tamoxifen as first-line adjuvant endocrine therapy in postmenopausal women.”

It took several years of teeth gnashing and committee pronouncements bsfore the breast cancer “intelligencia” finally agreed, but currently, the sentiment is strong enough that most docs reject a first opinion of tamoxifen, at least in this case.

For me as a patient, the AI is a no-brainer, and I am also not sold on AC as the best chemo option, as I am particularly struck by Dennis Slamon’s recent comments on several of our programs about his belief that anthracyclines no longer have a role in the adjuvant breast cancer setting — not in HER2-positive or HER2-negative disease, regardless of nodal status.

Not many other investigators take such a strong stance, but Dennis’s track record and his slide set convinced me. So I’m back to TC, this time with five years or maybe even more of an AI, depending on how I tolerate it.

This case raises the issue of adjuvant trastuzumab without chemotherapy. Everyone knows we don’t have definitive randomized data on this important clinical question and are stuck with indirect comparisons and laboratory predictions. Most docs prefer adding some type of chemo to trastuzumab unless it’s just too risky.

The question is, how old is too old? Or maybe, how comorbid is too comorbid? Given the high rate of early relapse in both ER-negative and HER2-positive disease, patients without major medical problems might need to be in their nineties to avoid a recommendation for a chemo/trastuzumab cocktail.

As for me — at age 70 and in otherwise good health, G–d willing — I’m going with a taxane alone with trastuzumab. I might even be tempted to take a shot at weekly nab paclitaxel/trastuzumab, although the thought of Cremophor^® paclitaxel and the thought of my manic self on corticosteroids might be worth the price of admission to friends, family and colleagues, as there is a pretty good chance I’d find myself down at the local Sizzler just before closing time, devouring everything on the pasta bar while my hypothalamus was driven insane by a cortisol bath.

In reviewing these and other findings in the enclosed survey, and particularly trying to put myself in the place of people facing these vexing situations, one other thought is relevant, namely that patients themselves might find these types of data interesting and useful. Most people don’t seek second, third or tenth opinions, but surveys like this might serve that function by providing a snapshot of the variability that currently exists in clinical oncology practice.

Neil Love, MD
DrNeilLove@ResearchToPractice.com
December 6, 2007

Select Publications