Are There Subsets of Patients Who Should or Should Not Receive an Anthracycline?
Amplified HER2/neu
Hazard Rates for Recurrence of Anthracycline vs Non-Anthracycline Regimens in a Meta-Analysis of Four Randomized Trials, Using Subsets Defined by Molecular Profiles[18]
The use of biomarkers as a method for identifying patients who might not benefit from a particular drug began in earnest after the observation in a Cancer and Leukemia Group B (CALGB) adjuvant chemotherapy trial that dose escalation of doxorubicin(Drug information on doxorubicin) benefited only those whose tumors overexpressed HER2/neu receptors.[16] Subsequently, it has been shown in other studies and in meta-analyses of trials comparing anthracycline regimens with non-anthracycline regimens that an anthracycline regimen is superior to CMF regimens primarily in patients whose tumors have amplified or overexpressed HER2/neu.[17, 18] In one of these meta-analyses, the hazard for recurrence was decreased by 30% (significant) for those with amplification of HER2/neu—but only by a modest 10% (not significant [NS]) among those with a normal HER2/neu (Table 6). Another meta-analysis identified 3 studies, including the CALGB trial described above, in which more intensive (either higher or more frequently administered) doses of anthracycline and cyclophosphamide(Drug information on cyclophosphamide) (and in one study, 5-fluorouracil as well) were compared with more conventional dosing schedules of these drugs.[17] The more intensive regimens resulted in a 46% reduction in the hazard of recurrence among those with HER2-positive tumors but no reduction in those with HER2-negative tumors. However, there were other differences beside doxorubicin dose or dose intensity in the regimens included in these studies, and it is plausible that the observed differences were not simply a result of different doses of anthracycline. In CALGB 9344, simple escalation of the doxorubicin dose from 60 mg/m2 to either 75 mg/m2 or 90mg/m2, with a constant dose of cyclophosphamide, did not improve disease-free or overall survival in patients with either HER2-positive or HER2-negative tumors.[19]
A scientific rationale for this interaction between anthracyclines and HER2/neu amplification or overexpression has not been established, and the predictive value of HER2/neu is not limited to the anthracyclines. A similar correlation exists with the use of taxanes. In CALGB trial 9344, patients were randomly assigned not only to receive different doses of doxorubicin but also to receive either paclitaxel(Drug information on paclitaxel) or no paclitaxel following 4 cycles of doxorubicin and cyclophosphamide. Paclitaxel significantly reduced the hazard of recurrence by 50% in those with HER2-positive tumors but reduced the hazard by only 16% (NS) in those with HER2-negative tumors.[19] In a small meta-analysis of three studies (one of which was CALGB 9344) comparing a taxane in one arm to no taxane in the other and for which the HER2 status of the patients was reported, the hazard of recurrence was reduced by 40% among those with HER2-positive tumors and by only 17% in those with HER2-negative tumors.[17]
From these studies we can conclude that patients with HER2-positive tumors derive greater benefit than other patients from either an anthracycline or a taxane, and they likely derive greater benefit from more intensive chemotherapy regimens as well. However, this may be equally true for cyclophosphamide, 5-fluorouracil, and any other drug we use in the adjuvant setting.
Amplified topoisomerase IIα
One, but not the only, mechanism of anthracycline anti-tumor action is suppression of topoisomerase IIα (TOP2A), an enzyme important in the repair of double-strand DNA breaks that occur after treatment with some cytotoxics. The gene for TOP2A is located very close to the gene for HER2/neu on chromosome 17, and it is often amplified when HER2 is amplified. TOP2A amplification is uncommon in tumors without HER2 amplification; however, not all tumors with HER2 amplification have TOP2A amplification as well. There are also problems in accurately identifying tumors with TOP2A amplification. In the studies in Table 6, 123 of the samples were assessed for HER2 and TOP2A using fluorescence in-situ hybridization (FISH) in both a local and a central laboratory, and the results were compared.[18] Only 6% of the HER2 results but 31% of the TOP2A results were discordant. Another study concluded that FISH assays lead to an overestimation of the number of tumors with amplified TOP2A because the commercially available FISH probes hybridize to more than the TOP2A gene.[20] These investigators found that 9 of 40 specimens they assessed using FISH were co-amplified for HER2 and TOP2A, but that only 2 of these 9 specimens (22%) were found to be amplified for TOP2A when assessed by high resolution representational oligonucleotide microarray analysis (ROMA).
In spite of these problems with defining this group of tumors, it makes sense that tumors with amplification of TOP2A might be particularly sensitive to anthracyclines, and there is considerable evidence to demonstrate that this is true. In the four-trial meta-analysis shown in Table 6, the reduction in the hazard of recurrence was 39% for patients with TOP2A amplification and only 11% (NS) for patients with normal TOP2A.[18] However, almost as many patients had deletion of TOP2A as had amplification, and in this group the hazard rate was reduced by 45% as a result of anthracycline treatment! This is not explained by the working hypothesis that tumors with excessive TOP2A will be particularly sensitive to inhibition by anthracyclines.
Further evidence for the possible importance of TOP2A as a predictor of response to anthracyclines comes from an as yet unpublished analysis of BCIRG 006 (see above).[14] All tumors in this trial were HER2-positive, and 35% of these tumors were co-amplified for TOP2A. The addition of trastuzumab(Drug information on trastuzumab) to ACD significantly improved the disease-free survival of patients whose tumors were not co-amplified for TOP2A (5-year survival, 70% vs 80% for ACD and ACD-H, respectively; P < .001). However, for patients with co-amplification of TOP2A there was no advantage from adding trastuzumab (5-year survival, 83% vs 85% for ACD and ACD-H, respectively; P = .60). This suggests that the doxorubicin in the ACD regimen is so effective for patients with co-amplified HER2 and TOP2A that there is no advantage from using trastuzumab as well. Alternatively, one might argue that the use of a non-anthracycline combination (such as DCarboH in the BCIRG 006 study) is as effective as and less toxic than the anthracycline regimen (5-year survival, 83% vs 82% for ACD and DCarboH, respectively; P = .30).[21]
As provocative as these observations on TOP2A are, it is too early to use them in standard practice. There is not agreement on the way tumors are defined as being TOP2A-amplified. The important factor may be expression rather than amplification of TOP2A, and, unlike with HER2, there is poor correlation between gene copy number and gene product. Almost all of the evidence is retrospective, and the analysis of the BCIRG 006 study has not yet been published in a peer-reviewed journal. None of the results have been confirmed in an independent, prospective randomized trial, and the results from trials reporting a correlation between TOP2A and anthracycline treatment are inconsistent. Some large trials and meta-analyses of these studies have failed to show any correlation at all.[21]
Triple-negative and BRCA1-associated cancers
Patients whose tumors are negative for estrogen receptors (ER), progesterone(Drug information on progesterone) receptors (PR), and HER2 have a worse prognosis than any other groups of patients. However, this entity is not easily defined by a single set of molecular characteristics. Most, but not all, cancers defined as basal-like in tissue microarrays are triple negative, and most, but not all, triple-negative cancers are basal-like.[22] TOP2A is regularly amplified in one subgroup of basal tumors, but triple-negative tumors may overexpress TOP2A without amplification.[18] Many, if not most, triple-negative tumors have loss of BRCA1 function even though the gene is rarely mutated the way it is in patients with heritable breast cancer.[22] One function of BRCA1 is repair of double-strand DNA breaks. Thus, tumors without BRCA1 or BRCA1 function are likely to be particularly susceptible to drugs, such as the anthracyclines or platinum agents, that induce double-strand DNA breaks—or to drugs, such as the poly(adenosine-disposphate-ribose) polymerase (PARP) inhibitors, that inhibit alternative mechanisms for repairing DNA breaks. In vitro, tumor cells with a mutant BRCA1 are relatively more resistant to spindle poisons, such as paclitaxel and vinorelbine, than to topoisomerase inhibitors such as etoposide(Drug information on etoposide).[23]
In two retrospective analyses, a higher response to anthracycline-containing neoadjuvant regimens was seen among tumors that were triple negative than among those with other molecular profiles.[24, 25] Similar results were obtained in a prospective trial of neoadjuvant doxorubicin and cyclophosphamide followed by paclitaxel in 144 women.[26] In the meta-analysis shown in Table 6, the advantage from an anthracycline in patients with triple-negative tumors (23% reduction in hazard of recurrence) is almost as great as that seen in patients with HER2-positive tumors (33% reduction).[18] However, drugs other than anthracyclines were included in the regimens used in these studies, so they do not prove that anthracyclines, taxanes, or any other form of cytotoxic treatment is more or less effective against triple-negative breast cancer.
The results from a number of other studies are inconsistent with these results. In one of the trials included in the meta-analysis shown in Table 6, CEF treatment resulted in inferior overall survival compared with CMF in 70 patients whose tumors had a core basal phenotype; the two regimens were equally effective in 29 patients with triple-negative, non-basal tumors.[27] In BCIRG trial 001, patients were randomly assigned to treatment with either TAC (docetaxel, doxorubicin, and cyclophosphamide) or FAC (5-fluorouracil, doxorubicin, and cyclophosphamide).[28] Among the patients with triple-negative tumors, there was a marginally significant (P = .051) advantage for TAC, with a 3-year survival of 74% vs 60% for TAC and FAC, respectively. Similarly, in the PACS 01 trial comparing 6 cycles of FEC with 3 cycles of FEC followed by 3 cycles of docetaxel(Drug information on docetaxel), there was a significantly better metastasis-free survival and overall survival for the taxane regimen than for FEC alone among patients with a basal-like profile; however, there was no significant difference in outcome among patients with a luminal pattern.[29]
Anthracyclines are clearly effective in triple-negative breast cancer, but the available data are insufficient to conclude that they should be used preferentially in this group of patients, especially since “triple negative” is not really a homogenous category but rather a composite of several different molecular subtypes for which the optimal therapy may differ.
Conclusions and Recommendations
Treat the most aggressive breast cancer with the most effective hemotherapies
Considerable evidence has been published over the past 15 to 20 years demonstrating that the proportional benefits from any and all cytotoxic therapies are greater in patients with hormone receptor–negative tumors. This is not merely a matter of a smaller absolute benefit in patients at lower risk of recurrence; cytotoxic drugs are more effective against receptor-negative tumors. This is true for CMF regimens[1] as well as for anthracycline and taxane combinations.[17, 18, 30] Newer molecular markers may be merely refining this process (Table 6). The evidence is still insufficiently robust to justify choosing among cytotoxics on the basis of a molecular profile.
. . . and consider “no chemotherapy” for the rest
The same studies have shown that the benefits of cytotoxic therapy are very small or even non-existent in the remaining patients, ie, those with a low risk of recurrence as determined by estrogen receptor status along with the newer molecular markers. For these patients, the toxicity associated with the most effective regimens may not be justified by the meager improvement in progression-free or overall survival. A common response to the poor therapeutic index of adjuvant chemotherapy in these low-risk patients is to use less toxic treatments, such as AC and now the non-anthracycline regimens such as DC. In doing so, however, it is important to recognize that the benefits from these treatments may be even smaller than the small benefits seen with standard regimens. Four cycles of AC is clearly not the most effective regimen in common use for the treatment of early breast cancer. It is likely inferior to other anthracycline regimens and has been proven inferior to taxane combinations.
Adjuvant trials that provide no information on the molecular profile of the patients enrolled on the study cannot be considered definitive. In US Oncology trial 9733, a sample of 170 tumors was evaluated for HER2 status. Unfortunately, the specimens were not randomly selected from all patients enrolled in the trial but were taken from “. . . those patients who had relapsed as of December 2005 as well as cases available at Baylor University Medical Center (in Dallas), the primary hospital for several of the coauthors.”[8] Thus, the tumors selected were biased towards those with an early relapse. The percentage of tumors that were positive for HER2 was 21% for the AC arm of the study and 34% for the DC arm. The authors report that DC was superior (but not significantly) to AC in both the HER2-positive and HER2-negative patient subsets, but because of the way the samples were selected, this cannot be accepted as valid. If in fact there was a significantly greater number of HER2-positive tumors in the DC arm, this could account for some differences in outcome, since HER2-positive tumors seem to derive the greatest value from treatment.
As new trial results appear in the literature, be cautious when:
• A trial designed to show that one regimen is superior to another fails to meet its endpoint and the investigators conclude the two regimens are equally effective. One of two apparently equivalent regimens may still be inferior.
• A molecular marker has been shown to identify a subset of patients who will benefit from a particular regimen. All regimens may be ineffective in the group without the marker.
Financial Disclosure: The author has served as a consultant for Genentech (although not for any agent mentioned in this article).
