This retrospective cohort chart review study conducted in France assesses the long-term outcomes of patients with HER2+ eBC treated with adjuvant chemotherapy and trastuzumab, in a RW setting. The number of cases as well as the representativeness of participating centers, provides a unique dataset for evaluating the long-term efficacy of adjuvant trastuzumab.
Among the 11,185 patients who were treated with adjuvant trastuzumab in 2010 in France, it is difficult to precisely evaluate the percentage of eligible patients that were included in the final data base (n=2,311 cases) because the French medicalized information system program (PMSI) does not distinguish the categories of patients excluded or removed from the analysis (women treated with neo-adjuvant trastuzumab, men with HER2+ eBC, patients who did not received the standard duration treatment with trastuzumab). While the differences in eligibility criteria between patients in this study and in phase III interventional trials, assessing adjuvant trastuzumab such as HERA and APHINITY, excluded direct comparison [4,10], the general demographic and clinical characteristics of the patients in our cohort and these studies appeared very close, except for a higher rate of older patients (aged 60 years or older) in our cohort, i.e. 35% compared to 16% in the HERA trial, and 20% aged 65 years or older versus 13% in the APHINITY trial [4,10]. These differences reflect the exclusion of older patients from clinical trial as well as the propensity of physicians to propose participation in clinical trials to few elderly patients even if the protocol does not exclude them. The difference in the proportion of patients aged 60 years or over explains that the percentage of postmenopausal patients was higher in our cohort (56%) than in interventional trials where this criterion was described (i.e. 45% in the HERA trial). The ratio HR+/HR- observed (63/37) is close to 60/40 usually found in an incident breast cancer population, and notably in the APHINITY trial (64/36) [4,10].
In France, the usual recommendation in real-world practices is to regularly follow-up patients with disease-free eBC during a period of 5 years, and then to space out or even discontinue the follow-up. The impact of this recommendation can be observed in the rate of patients who no longer consult at 5 years after the initiation of trastuzumab. The overall cumulative proportion of these patients is 6.6% (7.7% based on disease-free patients) but increases to 14.4% and 17.4% respectively at 6 years.
Although the number of patients with disease-free eBC at least 6 years post-trastuzumab remains high, the significant drop at 5 years follow-up time may impact the population still documented over this period (less N+ patients lost to follow-up at 6 years, 11.8% versus N- 17.2%). This justifies reporting 5 years follow-up even if the results at 7 years remain relevant for each subgroup.
The 5- year and 7-year DFS in the overall population was 85.4% and 76.3% respectively but it varied significantly according to hormonal status and nodal involvement. The same observation can be made for both the 5-year MFS (89.5%) and the 7-year MFS (84.2%) which are respectively 83.4% and 77.7% in the HR- subgroup and conversely 83.5% and 74.9% in the N+ subgroup.
In the N+ subgroup, patients with 4 or more positive nodes have lower risk of recurrence: 64.4% at 5 years and 50% at 7 years for the DFS, 70.5% at 5 years and 58.7% at 7 years for the MFS.
The evaluation of DFS and MFS at 5 years in HER2+ eBC is of real interest, because most clinical trials limit the observation period to 3 or 4 years of follow-up.
Due to the differences in the characteristics of the population included, patient management, follow-up process and time elapsed until assessment, comparisons of the DFS and MFS rates of this study with those of other studies in a similar population must be interpreted with caution. The final analysis of the HERA study at 11 year-follow-up showed that the addition of 1 year of adjuvant trastuzumab resulted in a constant 24% relative reduction in the risk of DFS event [8]. In the PHARE trial, a large phase III multicenter randomized French clinical trial assessing 6 months versus 12 months of adjuvant trastuzumab for patients with HER2+ eBC, the 3-year MFS (which equals to 3 years and 4 months in the present study) in the N+ subgroup (88.9%) was very close to those observed in this study at the same time point (90.7%). Beyond the fact that the 3-year MFS in the PHARE trial confirm our result, it shows that during the 4thand 5th year following post trastuzumab treatment, an additional 5% of N+ patients show a metastatic recurrence [12]. The trend pointed out by the Kaplan-Meier analysis seems to indicate a similar tendency in the 6th and 7th year.
The classification of the N+ subgroup (1-3 positive nodes and 4 or more positive nodes) is also particularly interesting. At 7-year follow-up, the DFS of patients with 1-3 positive nodes was 73.0% and the MFS 80.8%, while for patients with 4 or more positive nodes this was 50.0% and 58.7% respectively, showing 41.3% of metastatic recurrence.
Regarding the hormonal status, the differences between the 2 subgroups were less pronounced, but remained significant (p < 0.001) and confirmed by the Cox model: the 7-year DFS was 69.2% (30.8% of cumulative risk of event) and the 7-year MFS was 77.7% for HR- patients (versus 80.5% and 88.0% respectively in the HR+ subgroup).
Indirectly, these figures point out the difference in the nature of the recurrence for the different clinical profiles. For N+ patients most of recurrences were metastatic, especially in cases involving 4 or more positive nodes, whereas in the N- subgroup, the main recurrences were non-metastatic.
Overall, among patients who developed distant metastasis, the sites of involvement were visceral, mainly pulmonary and/or liver (73.8%), bone (52.6%) and brain (41.8%). There was a higher percentage of brain metastases in the HR- subgroup than in the HR+, and more in the N+ subgroup than in the N-. These differences have to be interpreted with caution, because in the N- subgroup the rate of distant recurrence was the lowest.
The limitations of this retrospective cohort study are inherent to the design of all RW studies, data interpretation should be taken cautiously. For instance, the rate of lost to follow-up patients after the end of the treatment may include both discontinuations based on patient decision and discontinuations decided by the sites for patients with no events. A comparison between the characteristics of the overall population and those of patients not lost to follow-up shows that the population not lost to follow-up at 8 years can be extrapolated to all patients and therefore that the cumulative rates of recurrence (whatever its nature) and metastatic recurrences observed at this time do not include any evaluation bias. Moreover, the comparison related to the N+ subgroup shows no difference between the distribution of the characteristics of all the patients included and that of the patients not lost to follow-up at 8 years. Although the slightly larger proportion of N+ patients not lost to follow-up at 8 years is likely to have a slight impact on the probability of recurrence for the entire population, there is no potential impact on the analysis on subgroups.
Since the publication of the results of the three phases III (NSABP B-31, NCCTG N9831, HERA) a decade ago, which provided solid evidence that adding trastuzumab to adjuvant treatment was indisputably increasing the PFS, time to first distant recurrence and OS in patients with HER2+ eBC, the prognostic paradigm has definitively changed for this population [4-7]. HER2+ eBC, remains associated with aggressive behavior in breast cancer, however, targeted therapy can significantly improve the outcomes of the disease and alleviate the burden associated with this characteristic.
The results of this study demonstrate less optimistic trends. In the HR+ and N- subgroups it is true that the 7-year probability of a recurrence (respectively 19.5% and 12.8%) and distant recurrence (12.0% and 5.3) is relatively low but in the two other subgroups (HR- and N+) the outcomes are not the same. The 7-year probability of a recurrence is 30.8% for HR- and 33.2% for N+ subgroups, with a probability of distant recurrence of 22.3% and 25.1% respectively. Both quantitatively, by the frequency of events, and qualitatively, by the nature of the recurrence observed, the HR- and the N+ are two subgroups for whom the 7-year outcomes are relatively poor.