Introduction
Colorectal cancer (CRC) is the third most common cancer in men (746,000 cases, 10.0% of the total) and the second most common cancer in women (614,000 cases, 9.2% of the total) (Ferlay et al.
2014) worldwide. CRC is the second most common cancer in Europe, accounting for 13.0% of all cancers apart from non-melanoma skin cancers. In 2012, there were 214,866 deaths from CRC in Europe (12.2% of the total number of cancer deaths and the second most common cause of cancer-related deaths) (Ferlay et al.
2015).
The incidence and mortality of CRC increase with age. Approximately, 67% of CRC patients are aged 65 years and older, and the mortality in this age group is 77% (Ferlay et al.
2015). The age of the European population is increasing, with those aged 65 + years estimated to increase from 93.2 million in 2013 to 124.8 million in 2030 (European Commission et al.
2014). Therefore, a considerable increase in the burden of CRC is expected.
Elderly patients (≥ 65 years of age) are underrepresented in clinical trials (Denson and Mahipal
2014; Hutchins et al.
1999; Talarico et al.
2004), including CRC trials (Dotan et al.
2012; Schiphorst et al.
2014). This leads to a lack of evidence regarding the effectiveness of treatment for this population, resulting in heterogeneous recommendations in treatment guidelines for patients with CRC (Dotan et al.
2012; Kaźmierska
2012).
In Europe, approximately 25% of patients with CRC present with the metastatic stage at initial diagnosis, and approximately 50% of CRC patients will develop metastases (Van Cutsem et al.
2014). The ‘most common’ standard of care for first-line therapy consists of adding targeted therapies to chemotherapy agents (with or without tumour resection) (Stein and Bokemeyer
2014; Kelly et al.
2014).
There is a substantial difference in survival across CRC stages (stage I, 97.4% at 5 years; stage IV: 7.5% at 5 years). However, survival estimates present fewer differences between age groups regardless of the stage, including 63.1% at 5 years in the 40–49 years age range, and 58.3% at 5 years in the 70–79 years age range (Cancer Research UK
2015a,
b). The elderly might achieve similar survival rates when they are appropriately treated. A meta-analysis of four clinical trials showed that the ≥ 70-year-old group benefits from chemotherapy similar to younger patients (Folprecht et al.
2006). Nevertheless, treatment patterns and guideline recommendations for metastatic CRC (mCRC) vary, especially for the elderly (Dotan et al.
2012; Kelly et al.
2014; Shenoy and Harugeri
2015).
Observational studies might play an important role in providing data regarding the influence of age on the effectiveness of first-line treatment in mCRC in real-world practice (Concato
2012). A systematic review of observational studies was conducted to evaluate whether the effectiveness of the standard of care used as first-line therapy (as targeted therapies and/or chemotherapy regimens) for mCRC differs by age.
Discussion
This systematic review of the influence of age on survival outcomes in mCRC patients included observational studies that reflect the standard of care in mCRC patient management in real-world practice. A comprehensive search of recently published studies was conducted. All included studies were considered of acceptable or high quality. As a strength, the median duration of follow-up was 22.8 months, which is longer than that reported in clinical trials (18 months) (Lieu et al.
2014).
The focus of the review was on the age cutoff of 70 years, which is the cutoff value most frequently used in studies to define the elderly. In addition, those aged 65–75 years vs. those aged ≥ 75 years were compared. Such a comparison excludes young patients and uses a narrower interval to define the elderly, providing a more accurate picture of the differences among those over 65 years of age receiving the standard of care.
The results suggest that age did not have an effect on PFS and held for all age groups and for the pooled median of PFS using the cutoff value of 70 years. Only one study showed a significantly lower PFS in patients ≥ 75 years of age than in those < 75 years of age (Hofheinz et al.
2014). However, in the same study, PFS did not differ by age when the cutoff value was 70 years. In this study, there appears to be a shift towards monotherapy in the age group from 70 to 75 years, as reflected by the difference in the proportion of monotherapy between the two age groups (≥ 70 years: 21.7%, ≥ 75 years: 31.7%).
OS was the most frequently reported survival outcome (9 out of 11 studies). In most of the studies (
n = 6,
N = 6760), the elderly had a lower median OS than the younger age group, regardless of the age cutoff value used (70 years; 65–75 vs. ≥ 75 years). This difference was statistically significant in three studies (Kozloff et al.
2011; Hofheinz et al.
2014; Rouyer et al.
2016). The pooled analysis showed consistent results by age, and the median summary was slightly worse for those ≥ 70 years than for those < 70 years (≥ 70 years: 22.86 months; < 70 years: 27.04 months). These results were consistent with those of the fourth study that we could not combine in our summary analysis (< 70 years: 25.8 months, ≥ 70 years: 22.7 months,
P < 0.0008) (Hofheinz et al.
2014).
The proportion of patients with synchronous metastasis was similar across the different age groups (Kozloff et al.
2010; Slavicek et al.
2014; Parakh et al.
2015). Nevertheless, one study (Slavicek et al.
2014) showed that the risk of death increased by 25% for synchronous patients in comparison with metachronous patients. In the same study, the risk of disease progression increased by 13% for synchronous metastasis patients. Both results were regardless of age. Furthermore, Parakh et al. (
2015) reported that the rate of primary tumour resection (58% vs. 47% vs. 45%,
P = 0.037) and metastatic resection (26% vs. 21% vs. 6%,
P < 0.001) for synchronous metastasis patients declined significantly with increasing age (65–74, 75–84, and ≥ 85 years, respectively). This finding might partly explain the significant difference in OS results (65–74 years: 26 months; 75–84 years: 20 months; ≥ 85 years: 11 months,
P < 0.001) for that study, especially that the median OS of patients who underwent metastatic resection vs. those who did not undergo metastatic resection showed a longer OS for those with metastatic resection in the same age group (65–74 years: 50.4 months vs. 19.8 months, respectively, HR 0.20 (0.13–0.32),
P < 0.0001; 75–84 years: 37.8 months vs. 20.7 months, respectively, HR 0.26 (0.17–0.39),
P < 0.001; ≥ 85 years: 20.7 months vs. 10.4 months, respectively, HR 0.43 (0.13–1.36),
P = 0.15). In the same study, the use of CT declined with increasing age (65–74 years: 84%; 75–84 years: 69%; ≥ 85 years: 34%,
P < 0.001). Thus, this patient management practice may refer to a treatment practice of using less aggressive treatment with increasing age.
Only two studies showed better OS results in the elderly group than in the younger groups (Dirican et al.
2014; Fukuchi et al.
2013), although these results did not reach statistical significance, and the studies used different age cutoff values (65 and 75 years, respectively). In Fukuchi et al. (
2013), patients ≥ 75 years (
n = 18) had better OS than patients < 75 years (
n = 108), which might be partly explained by a small number of elderly cases. Both age groups received the same treatment (irinotecan-based CT + bevacizumab). In Dirican et al. (
2014), the elderly, defined as patients ≥ 65 years of age, had slightly better OS than those < 65 years of age (31 vs. 22 months, respectively). The design of this study was slightly different from the others; patients (regardless of age) were enrolled in two different cohorts (cohort A: patients were treated with chemotherapy in combination with bevacizumab; and cohort B: patients were treated with the same chemotherapy as cohort A but without bevacizumab). Then, survival results were re-analysed by age group. In both cohorts (with and without bevacizumab), elderly patients showed slightly better results than patients < 65 years of age in terms of OS (< 65 years: 22 months, ≥ 65 years: 31 months) and PFS (< 65 years: 9 months, ≥ 65 years: 11 months). However, in that study, elderly patients represented only 26.8% of cohort A (CT plus bevacizumab), while they represented 52% of cohort B (CT alone). This proportion may reflect the treatment practice for the elderly in comparison with that for younger age groups, especially since all included patients were receiving combination CT (FOLFOX, XELOX, XELIRI, or FOLFIRI regimens). This finding is consistent with the results of other studies (Hofheinz et al.
2014; Kozloff et al.
2010; Parakh et al.
2015; Sahm et al.
2016; Slavicek et al.
2014; Tahover et al.
2015), regardless of the age groups of comparison, in which the intensity of CT and the dose frequency of the targeted therapy decrease with increasing age. Elderly patients tend to receive less combination therapy (FOLFOX and FOLFIRI) and more monotherapy (5-fluorouracil and capecitabine). This treatment pattern was also observed in studies that compared two elderly age groups (65–75 years vs. ≥ 75 years). In those studies, the proportion of patients ≥ 75 years of age receiving monotherapy alone was approximately three times higher than that of patients 65–75 years of age, regardless of the targeted therapy used (Kozloff et al.
2010; Slavicek et al.
2014). Similarly, in another study that reported the use of triple CT backbone by age group (Kozloff et al.
2010), the proportion of patients treated with triple CT backbone dropped progressively with increasing age (< 65 years: 55.3%, 65–75 years: 48.6%, 75–80 years: 32.7%, and ≥ 80 years: 26.1%).
These results might reflect different treatment patterns in the elderly by oncologists, partly explained by a lower tolerance to the adverse events of aggressive CT, which are expected to be more in the elderly than in younger groups. This expectation seems not to be justified by the PS or number of metastatic sites, since the difference observed according to age group was small. The difference in treatment by age might have an influence on the lower OS observed in the elderly. The fact that younger people are expected to live longer might also influence these results.
ORR showed a lower response in the elderly than in younger groups regardless of the age cutoff, although it was statistically significant in only one study (Hofheinz et al.
2014). Fukuchi et al. (
2013) showed the opposite results (Fukuchi et al.
2013) with the elderly having a higher response than the younger age group, although this was not statistically significant. This result might be explained by the fact that in Fukuchi et al. (
2013), both age groups received the same treatment (irinotecan-based CT + bevacizumab), while in the rest of the studies, treatment varied. The small sample size in Fukuchi et al. (
2013), with only 18 patients over 75 years of age, might influence the results reported.
In this systematic review, some limitations should be noted. The search might have missed some studies, although the search strategy was comprehensive and included grey literature. Publication bias might have resulted from the fact that studies with positive results are more likely to be published. The patients’ baseline characteristics showed an imbalance according to age group. The elderly presented a higher frequency of comorbidities and less aggressive treatment, which might have influenced the results. Only 6 out of 11 studies reported the use of propensity score matching to adjust for these baseline differences (Dirican et al.
2014; Fourrier-Reglat et al.
2015; Fukuchi et al.
2013; Hofheinz et al.
2014; Sahm et al.
2016). The included studies did not consider the influence of the histological subtypes on tumour treatment (Hugen et al.
2014). Information on a major potential factor that might influence the outcome for OS was not available or reported in other age categories, as in the case of tumour location, which was analysed in three studies regardless of the age groups (Dirican et al.
2014; Slavicek et al.
2014; Tahover et al.
2015). The study by Slavicek et al. (
2014) was the only one to show a significant increasing risk for rectum over colon in both PFS and OS (PFS: rectum/colon, HR 1.10 (1.01–1.19),
P = 0.04; OS: rectum/colon, HR 1.13 (1.01–1.26),
P = 0.03). Another factor that impacts OS is the RAS/RAF mutation status. Although the included studies did not refer to the RAS/RAF mutation, we took into account that the biomarkers’ effect on targeted therapies was integrated into treatment guidelines after 2008 (Bellon et al.
2011; Carter et al.
2015). As a result of this, all metastatic patients of the included studies with cetuximab treatment were KRAS wild type (Fourrier-Reglat et al.
2015; Sahm et al.
2016). To reduce heterogeneity, studies with patients receiving only chemotherapy or untreated patients were excluded. Finally, the evaluation of PFS and ORR might differ between oncologists and studies.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.