Equations That Can Predict Risk of Treatment Failure

ABSTRACT: Three equations have been formulated to estimate the risk that men treated for clinically localized prostate cancer have extracapsular extension, lymph node involvement, and non-organ confined disease. The equation for extracapsular extension risk is based on pretreatment prostate-specific antigen (PSA) and Gleason scores. The equations for risk of lymph node involvement and non-organ confined disease also include T stage. The “beauty” of these type of equations is that they can be applied equally to surgically treated and radiotherapy treated patients. Physicians can use these models to advise individual patients and to stratify men into prospective randomized trials more accurately than using clinical stage alone. [Oncol News Int 6(Suppl 3):5-6, 1997]

Introduction

Three simple equations based on pretreatment prostate-specific antigen (PSA), T stage, and Gleason score can be used to predict path-ologic stage and the risk of treatment failure in patients with localized prostate cancer, reported Mack Roach III, MD, at the First Sonoma Conference on Prostate Cancer. Dr. Roach is Associate Professor, Radiation and Medical Oncology, at the University of California, San Francisco.

The equations are used to calculate risk of biochemical failure—rising PSA—following surgery or radiotherapy. Dr. Roach reviewed studies showing the close correlation between rising PSA and declining survival “to reinforce the point that PSA is likely to be a valid end point—surrogate end point, probably—for survival.”

The equation for extracapsular extension is based on pretreatment PSA and Gleason score (GS). The equations for lymph node involvement and non-organ confined disease add T stage to these factors (Table 1).

Risk for Extracapsular Extension

 The original equation predicted the risk for extracapsular extension (ECE). That equation is : ECE risk = (1.5 x PSA) + [(GS - 3) × 10].

For use in these equations, the nine possible Gleason scores are grouped into five major scoring categories: Gleason scores 5, 6, and 7 remain discrete categories, and scores of 4 or less and 8 or higher are lumped into a single category.

Applying the ECE risk to about 374 patients who underwent radical prostatectomy at UCSF, produced “an excellent correlation,” Dr. Roach said.

He next looked at how the equation would correlate with outcome of patients treated with radiotherapy. He found that the equation identified patients with relatively low, intermediate, and high risk of progression. “In general, it separated these patients out better than using other parameters, like T stage and Gleason score, although it’s unclear how much the equation adds over PSA alone,” Dr. Roach said.

“Certainly, one of the advantages to this approach is that it’s useful both in predicting pathologic stage, as well as the risk of failing radiotherapy.”

How Important Is Gleason Score?

 Reviewing longer-term data from 500 patients treated with radiation at UCSF between 1987 and 1995, Dr. Roach found the disease survival curve based on the extracapsular extension risk equation “does not look a whole lot different” than that based on PSA alone.

“I think that reflects a point that I’d like to make,” he said, “and that is, I don’t think that Gleason score is as important in patients undergoing radiotherapy as it is for patients undergoing surgery. I think that, as a prognostic factor, Gleason score is more important in patients undergoing radical prostatectomy.”

Dr. Roach presented data from several studies showing that for patients treated with radical prostatectomy, those with a Gleason score of 8 to 10 had about a 75% failure rate at 4 years. Furthermore, based on data presented by Lerner et al, patients with a Gleason score of 7 had a 40% failure rate; those with a Gleason score of 8 had a 50% failure rate, and those with a Gleason score of 9 had a 60% failure rate, even with pathologically organ-confined disease (Reference: Lerner SE et al: Risk Factors for Progression in Patients With Prostate Cancer Treated With Radical Prostatectomy. Semin UrolOncol 14:12-21, 1996).

Risk of Lymph Node Involvement

The equation for risk of lymph node involvement (N+) is: N+ risk = 2/3 PSA [(GS + TG - 8) × 10].

For this and the equation for non-organ confined disease:

• TG 1 = T1a and T1C, is assigned a value of 1
• TG 2 = T1b, T2a, is assigned a value of 2
• TG 2.5 = T2b, T2c, is   assigned a value of 2.5
• TG 3 = T3a, is   assigned a value of 3.

Comparing the calculations from the equation for risk of lymph node involvement with the actual experiences of patients treated at UCSF and at several other institutions (reviewed by Alan W. Partin, MD) showed “for most values, there is very little difference,” Dr. Roach said. “Although the equation tended to underestimate the risk in patients with high-grade disease.”

Non-Organ Confined Disease

The equation for risk of nonorgan confined disease (NOCD) is: NOCD risk = 3/2 PSA [(GS + TG - 4) × 10].

 “Basically, the correlation between the equation for predicting non-organ confined disease is very close to the data” in the multi-institutional study reported by Dr. Partin combining PSA, T stage, and Gleason score. “If you take that data and you apply it to patients who undergo radiotherapy, you get this nasty looking curve, which, in general, shows that patients at high risk of having non-organ confined disease” experience “a higher failure rate,” Dr. Roach said.

Incomplete Data Present Problems

One of the problems with currently available data and using it to try to develop a model, Dr. Roach acknowledged, is the lack of PSA data for patients who received radiotherapy a long time ago.

“Sometimes, what would happen is a patient would have radiation, say 3 years ago, and he’d be sort of lost to follow-up,” Dr. Roach said. “He’d show up at a clinic, and there would be a 2-year interval between PSAs, and the PSA would be 5.

“Now, in the literature, most people define PSA failure as the day that person comes in with an elevated PSA. But, in fact, the real failure time was some time earlier...The median time to failure, the disease-free survival, is longer for these people, creating an artifact.”

Dr. Roach said that the problem “is a sort of Catch-22 situation.” To make a good model predicting outcome for patients, you need lots of good information, but information is lacking on some of the very types of patients whose outcome the model is attempting to predict.

‘Beauty’ of Prognostic Factors

The beauty of using this type of equation in men with clinically localized prostate cancer to predict risk of developing extracapsular extension, lymph node involvement, and non-organ confined disease, said Dr. Roach, is that it can be applied equally well to surgically treated and radiotherapy patients. Knowing that a patient has a significant risk of failing radiation obligates us to “look for more,” he added.

The best options for these patients are using higher doses of radiation and hormonal manipulations based on their being identified as being at low, intermediate, or high risk of biochemical failure.