Special Feature

Lessons in Supportive Care, Part VI: Getting Our Experts to Agree

By Christopher E. Desch, MD, and Thomas J. Smith, MD

Case: You are the medical director for a newly formed oncology group, and final decisions about very expensive treatment rest with you. A 68-year-old woman comes to your office with a diagnosis of non-Hodgkin lymphoma and confusion. She is confused-not from the disease, but from her "expert" consultations. She was generally healthy, with a 10 cm peri-pancreatic mass and enlarged adjacent nodes. Performance status was 1. Lactic dehydrogenase (LDH) was 1500. Echocardiogram was normal a month ago. Her husband, an attorney, presents you with the consultations in Table 1.

The family is confused, distrustful, and frantic. They cannot explain the disparity among opinions. Of course, it is possible the family misinterpreted the facts, but they took careful notes. As the insurance companies dump tough decisions back to us (which is what capitation is all about) the buck must stop somewhere. It is on your desk.

Table 1

Expert Opinions

Expert Staging Treatment Cure rate given
Bone marrow
Chest CAT
oncologist at
famous center

MUGA scan
Gallium scan
CAT of whole

Would not
give CHOP

Does not discuss
prognosis. Suggests
that PC is more
effective but more toxic,
and that if it is given,
then high-dose chemo-
therapy is excluded
CAT of chest
no bone marrow

CHOP or Pro-
No intrathecal
40-50% Would
give high-dose chemo-
therapy if sensitive
Phone consult
famous real
lymphoma expert

Bone Marrow
CAT scans

Induction with
CHOP, then high-dose
chemotherapy. Intra-
thecal if bone marrow
25% with CHOP; 80%
with CHOP followed
by high-dose chemo-
Yours here


This should be a bread-and-butter, straightforward oncology case. After all, we have been doing randomized clinical trials in NHL for years and pride ourselves on our accomplishments. However, the wide variations in recommendations suggest that Smith's Rules of Oncology are in play. (See Table 2.)

Your group has agreed to live by consensus vote, and to use the best available data. That is, you will value high-power randomized clinical trials most, and use the most up-to-date information from predictive models, rather than guesses.

What is reasonable to know about prognosis?

Her prognosis is not good, but you do not know how bad it is, yet. The International Non-Hodgkin's Lymphoma Prognostic Factors Project (N Engl J Med 1993;329:987-994) would already give this patient bad marks for age greater than 60 years, high LDH, and possibly stage III. You still need to know about tumor stage (I, II vs III, IV) and the number of extra nodal sites. If the bone marrow and other CAT scans were normal, she could be "2, low intermediate risk," but it seems more likely that she is "3, high intermediate risk" or "4 or 5, high risk."


Smith's Rules of Oncology

1. If you cannot be effective, at least be trendy.

2. All decisions can be rationalized.

3. Each doctor is allowed one disease to treat irrationally.

The differences are big. If she is low risk, the five-year survival rate is 56%; for the others, 44%, 37%, and 21%, respectively. So, further information might be important to help her gauge her chance of survival.

What is reasonable to know about staging?

We stage to know about prognosis and to help choose therapy. And we can stop when we have enough information.

The patient should undergo further staging if she wants to know her prognosis. At a minimum, CAT scans and a bone marrow biopsy to assess the number of nodal and extra nodal sites would be useful. The wide variation in studies suggests that some of the consultants want more certainty than others, when it is not needed. An MUGA scan is not very likely to add important information to a normal echocardiogram. A Gallium scan may be useful in select cases; the National Cancer Center Network guidelines for lymphomas, presented earlier this year, suggest that there is still divided opinion on its usefulness. (Smith's Rule #1.) No big deal; get it if it helps determine therapy. (Smith's Rule #2.)

Now, how will you use the information to guide therapy?

What are the data on treatment?

Remember the Southwest and Eastern Cooperative Oncology Group trial of 899 patients, with 218 in each arm, showed no significant difference between CHOP and m-BACOD, ProMACE-CytaBOM, and MACOP-B (Fisher RI, et al. N Engl J Med 1993;328:1002-1006.) The overall survival rate was 52-54%. The toxic death rate was 1-6%, with CHOP being lowest.

Here is where all of Smith's Rules come into play. CHOP is simple in design, still toxic if done right (we have a 70-year-old in house now with WBC of 0.1 and platelets of 6,000), and costs one-quarter of some of the other regimens. It certainly does not have the panache of the newer regimens. (Smith's Rule #1.) Arguments about whether it is 4% better might be valid but are unprovable. (Smith's Rule #2.) And doctors who otherwise live and die by the randomized clinical trial sword will argue this one till the cows come home. (Smith's Rule #3). Again, it is probably no big deal as long as your oncologists agree to use one or, at most, two regimens by the book.

The variation begins with CHOP but does not end with it. At least two of the experts mentioned high dose chemotherapy as part of the initial treatment, one to state that ProMACE rules it out-for which there are no data or rationale. The other real expert suggested high-dose chemotherapy off-protocol when the randomized clinical trial data suggest no benefit. The randomized clinical trial of 464 patients with NHL in first remission showed no difference in DFS or OS. (Haioun C, et al. J Clin Oncol 1994;12:2543-2551.) The study of slow responders showed equivalent or better OS by continuing standard CHOP. (Verdonck LF, et al. N Engl J Med 1995; 332:1045-1051.) There is clear-cut benefit for young chemotherapy-sensitive NHL patients, better OS at five years (53% vs. 32%, P = 0.035), but 68 or beyond years would be far outside the available data both for toxicity and results. (Philip T, et al. N Engl J Med 1995; 333:1540-1545.) An objective observer would state, based on the evidence, that high-dose chemotherapy remains one unproven strategy in the management of elderly NHL patients.

One of the conservative oncologists was comfortable treating with limited data. She had made up her mind that CHOP, six cycles, was appropriate therapy. And if high-dose chemotherapy were to be considered in its proven setting-sensitive relapse-then all the staging tests could be done. Such a strategy would reduce testing by a considerable amount.

The real lymphoma expert has underestimated survival based on the international prognostic index or at least what we know about this patient. This expert has also dramatically overestimated the success of high-dose chemotherapy. He or she did not give figures for mortality for high-dose chemotherapy given to 68+ year olds. We do not know what it is, but it is not likely to be the same as for the 48-year-olds we have treated on and off clinical trials.

What are the consequences of expert disagreement?

These well-meaning but improbable interpretations of the data had major consequences for this patient and her family, and for you.

First, they now mistrust the health care system. And that means your job of "selling" them on one type of therapy is terribly, terribly hard. Who would not jump at the chance of an 80% long-term survival, promised them by the real lymphoma expert? Your job is to explain to them that the promised therapy has not been proven to be effective or safe.

Second, you have to muster the resources to do the staging and treatment. If you can treat-in good conscience-with six cycles of CHOP in the absence of bone marrow, Gallium and MUGA scan, and full body CAT scans, then you can save substantial amounts of resources for your other patients or for her. After all, if and when she relapses, you will repeat all those scans again.

Third, how will your treatment approach fit with the NCCN and other guidelines, yet to be published?

Fourth, the patient wants to know why not everyone believes the CHOP data, and one doctor "never" gives it, one "always" gives it, and one "sometimes" gives it.

Can we come to consensus?

Not left to our own devices.

In the past, we never had to come to consensus. There was enough "slack" in the system that doctors could do whatever they wanted with NHL, if they had staked it out as territory. (Smith's Rule #3.) We could give regimens that cost four times as much and produced more side effects. We could transplant chemotherapy-insensitive patients as a "last chance." Academic and other centers provide results to patients of protocols in progress, stating they are more effective when the appropriate trials have not been done.

Now, you as the oncology director or one of the oncologists, must agree to live under certain rules. Choosing what tests to do should be straightforward: Do them sequentially until you have enough data to make a prognosis and choose treatment. Then, choose one or, at most, two treatments (recognizing the indelible mark of the NCI on NHL treatment) and use only them. Define the role of high-dose chemotherapy based on current randomized clinical trials unless other compelling evidence is present: always for chemotherapy-sensitive relapse in young patients, otherwise only on clinical trial or under special circumstance. After treatment, do not do follow up tests-other than a careful history and physical-unless clinically indicated. And choose second-line regimens with the same critical eye as first-line ones, as some will bankrupt you. (Next topic: inpatients infusional vs. outpatient or bolus regimens.)

We can do much better explaining things to our patients. Not one of the experts asked the simple question: "What have the other doctors explained to you?" They each started with their own set of assumptions about staging and treatment, and never helped the family understand the very real differences in opinion. Not fact, opinion.

Final message

We can do better, and we must. Restricting second opinions seems heavy handed. But, asking experts to back up their opinions with data seems only fair; if they cannot, then they may come off your list of experts. And, asking your experts to explain their opinions to patients, including how and why they differ, makes good business and medical sense. (Dr. Desch is with the Massey Cancer Center, Medical College of Virginia-Virginia Commonwealth University.)