Ambrosone CB, Zirpoli GR, Hutson AD, et al. Dietary supplement use during chemotherapy and survival outcomes of patients with breast cancer enrolled in a cooperative group clinical trial (SWOG S0221). J Clin Oncol. 2019;JCO1901203. doi:10.1200/JCO.19.01203
To determine if an association exists between dietary supplement use during chemotherapy and outcomes for patients with breast cancer.
This is a prospective observational study. Supplement use was assessed using questionnaires issued to patients with breast cancer who were enrolled in a phase 3 chemotherapy clinical trial. Supplement use was evaluated through questionnaires applied twice, with the first given prior to patients' beginning chemotherapy (assessing prior supplement use) and the second given at planned completion of chemotherapy (approximately 6 months after treatment initiation).
Study participants included 1,134 adult breast cancer patients scheduled to receive either dose-dense or standard chemotherapy (cyclophosphamide, doxorubicin, and paclitaxel) as part of the investigator team’s primary randomized trial. This group represents those participants who completed 2 questionnaires on supplement use within the larger clinical trial (n=2,716).
Disease-free survival (DFS) and overall survival (OS)
In the 1,134 patients who completed both questionnaires, 251 recurrences and 181 deaths occurred. In this group, for the fully adjusted model, use of any antioxidant (grouped) both before and during chemotherapy was nonsignificantly associated with recurrence (adjusted hazard ratio [adjHR]: 1.41; 95% confidence interval [CI]: 0.98-2.04; P=0.06) and survival (adjHR: 1.40; 95% CI: 0.90-2.18; P=0.14). When analyzed in isolation, any antioxidant use only before or only during was not associated with worsening of either endpoint. The use of single antioxidants vitamin C, vitamin E, and coenzyme Q10 (CoQ10) did not have statistically significant associations. Use of vitamin A both before and during chemotherapy was associated with a significantly increased risk of recurrence (adjHR: 4.23; 95% CI: 1.32-13.57) and death (adjHR: 3.53; 95% CI: 1.05-11.93), but not carotenoids (significant increase in recurrence for use only during). Reported use of multivitamins and vitamin D either before or during chemotherapy or for both time points was not associated with survival outcomes.
Use of any antioxidants (defined by the author as including vitamin A, vitamin C, vitamin E, carotenoids, and CoQ10) both before and during chemotherapy was stated to be nonsignificantly associated with an increased risk of recurrence and death. Associations for the single antioxidants were not statistically significant, with the exception of vitamin A and carotenoids. The use of vitamin B12 or iron both before and during chemotherapy was associated with a statistically significant poorer disease-free survival (DFS). Multivitamin use was not associated with any difference in survival outcomes. Out of 1,430 patients who completed only the first questionnaire, 310 patients experienced recurrence and 222 patients died. These patients were more likely to be older and postmenopausal and have a higher body mass index (all P<0.05). They were also more likely to have poor prognostic factors such as estrogen receptor, progesterone receptor, or HER2/neu (human epidermal growth factor receptor 2) negativity, and 4 or more lymph nodes involved (P<0.01).
Commentary and Practice Implications
In the continued effort to ensure the safety of natural health products (NHPs) consumed by the public, especially in the context of cancer, the publication by Ambrosone et al1 offers additional information for review and commentary. This study has garnered media attention emphasizing potential for harm in using dietary supplements, including antioxidants, during chemotherapy. This has fostered understandable concern from people living with cancer and the healthcare practitioners overseeing their care. A close review of this study allows for interesting findings to be highlighted, while also revealing certain limitations, of which most are expected for a study of this design (observational with post hoc features). This commentary attempts to contextualize pertinent study findings, review the study’s strengths and limitations, and explore implications for public safety and clinical practice. In terms of practice implications, this study suggests caution. It also suggests patients should work with providers trained in the use of natural therapies before implementing them—ideally in coordination with their oncology team for best safety.
While concern about the use of antioxidants concurrently with chemotherapy is justified by some reports of worse outcomes,2 the totality of available research reveals an absence of any definitive conclusions, as there are, conversely, studies finding possible improved outcomes and reduced side-effect burden.3 In some cases, the literature shows that supplementation with certain products, including some mentioned in this study, is beneficial and indicated during administration of certain chemotherapy drugs. For instance, while vitamin B12 was associated with possible worse outcomes with the specific chemotherapy drugs given in this study, it has been found to be safe and beneficial when given with pemetrexed (a drug used for lung cancer).4 Overall, as presented in a 2015 systematic review, the effects of antioxidant supplementation during cancer therapy remain unclear, requiring further exploration.5
The study’s main conclusion, which suggests that the use of all antioxidants during chemotherapy may be harmful, suffers from numerous limitations and should be reframed and revised.1 A major limitation of this study is the apparent post hoc approach adopted for selecting certain antioxidants for review; the selection appears arbitrary, as no clear rationale is provided for why certain agents were chosen or presented, while others were not (eg, melatonin, an antioxidant, is not mentioned). This approach allows for the risk of data dredging, as it appears only certain antioxidants were selected for analysis after review of multiple potential options and after data was collected, resulting in the possibility of spurious results due to multiple testing. While this primary limitation does not warrant dismissal of the study, it does bring forth a flaw that should have been noted, and a more conservative statistical correction for multiple analyses should have been subsequently applied. The authors also note that the number of patients actually taking antioxidants was small, reducing the statistical power and potentially resulting in false associations.
In the analysis, fully adjusted hazard ratios (where important variables were controlled by investigators, including smoking and body mass index) for both DFS and OS did not reveal any significant associations for any time measure, other than for vitamin A and carotenoids specifically.
A close review of the analyses regarding antioxidant use reveals additional concerns. When the researchers looked at any antioxidant use, for both DFS and OS, they found no significant associations (for either only before; only during; or before and during), with “only during” use actually showing a nonsignificant reduction in DFS (adjHR, 0.93; CI, 0.52-1.64). It should not, however, be overlooked that use of any antioxidant both before and during almost reached a significant association for poorer DFS, indicating that perhaps long-term use (prior use which then carries over into chemotherapy) could be problematic (P=0.06). This may have been due to a small sample size, which was not conducive to achieving statistical significance. Further research is required to determine true risk.
Taken together, this work by Ambrosone et al provides a good effort to address important questions and does reinforce the concept that just because something is natural does not necessarily mean it is safe.
Regarding specific antioxidants, vitamin A and carotenoid use appeared to worsen outcomes, which is consistent with the literature for these agents and cancer risk.6-9
Adjusted analysis for DFS and OS for the majority of non-antioxidant supplements showed no association for either endpoint, with the exception of iron, vitamin B12, and omega-3 fatty acids.
Vitamin B12 and Iron
Both iron and vitamin B12 were found to be significantly associated with worse DFS and OS. Regarding vitamin B12, use both before and after chemotherapy was associated with significantly worse DFS (adjHR: 1.83; CI: 1.15-2.92; P<0.01) and OS (adjHR: 2.04; CI: 1.22-3.40; P<0.01). No significant associations were found for use only during or only before chemotherapy. Iron use only during chemotherapy was significantly associated with worse DFS (adjHR: 1.79; CI: 1.20-2.67; P<0.01) and OS (adjHR: 1.71; CI: 1.06-2.76). Iron use both before and after chemotherapy was associated with a nonsignificant worse DFS (adjHR: 1.91; CI: 0.98-3.70; P=0.06). These findings are consistent with emerging research that vitamin B12 supplementation may be associated with increased cancer risk, specifically lung cancer,10 and that iron intake/status may be associated with breast cancer11; however, that has not been investigated with chemotherapy treatment. The significant associations identified by Ambrosone et al add to the literature indicating that caution is warranted when considering supplements such as vitamin B12 and iron, as they may be detrimental.
Omega-3 fatty acids
The use of omega-3 fatty acids both before and during chemotherapy was associated with worse DFS (adjHR: 1.67; CI: 1.12-2.49), but no significant association was found when used only during (adjHR: 1.15; CI: 0.62-2.13). No significant associations were found for OS. A limitation to this component of the study is the fact that the questionnaires’ query on omega-3 fatty acid use included a wide mix of sources with variable potential effects on outcome.
The major limitation of the available relevant research is that studies often have small sample sizes and are of short duration, where long-term effects of supplement coadministration with chemotherapy may not be realized. Ambrosone et al1 explored this area and presented analyses of participant-reported antioxidant use with regard to DFS and OS in a relatively large group (n=1,134). It is notable that the authors do mention that compared to other estimates of antioxidant use by patients, there was a smaller percentage of participants reporting supplement use within the sample size of this study.
Observational research allows us to identify associations within a population where the investigators are not directly controlling variables (compared to clinical trials where there is specific control over what participants are exposed to). A classic example of this form of research is the work done on cigarette smoking, where investigators observed large populations of smokers to see the effects, without giving the public cigarettes to see what would happen. It is important to note that it is not possible to determine causality from this form of research.
It is of note that only a small portion of the total patient population used supplements. For example, only 138 patients (12.2%) reported using oral vitamin C during chemotherapy. It is important to keep in mind the fairly small sample size from which the author’s conclusions stem, ranging between 5 to 551 people assessed based on use of each supplement (this is a limitation aptly presented by the study authors).
Overall, this observational study does add information to help guide safer use of health products around chemotherapy, specifically antioxidants; however, this is offset by major unaddressed limitations. Based on the present research, it appears that the safety of any antioxidant use during treatment, especially long-term use that carries over into treatment, is unclear and warrants further consideration. This study gives credence to the potential that carotenoids and vitamin A may worsen outcomes when combined with the chemotherapy agents in this trial, meriting cautious avoidance.
Further, the questionnaires in this study did not allow patients to account for dietary intake of supplements, such as through enriched nutritional beverages or protein powders, which can include additional antioxidants. However, the study did adjust for patients’ reports of taking 5 or more supplements.
The study published by Ambrosone et al offers some valuable information for clinicians and new insights into areas of research that warrant further investigation. The major strength of this observational study, setting it apart from others, is the fact that it was conducted within the context of a clinical trial and assessed both prior supplement use and use during chemotherapy. This design reduces, albeit not completely, the potential for important recall bias by participants.
As with the majority of observational studies, it is difficult to tease apart the associations between specific isolated variables (in this case, supplements) and specific endpoints. Is the specific supplement itself causing worse outcomes or is the supplement a marker for another factor? For example, vitamin B12 is only associated with worse DFS when used both before and during chemotherapy (not during alone), possibly indicating that people who would need B12 supplementation may have had longstanding poorer health (eg, malnutrition requiring a supplement). Adding to our interpretation of these findings, a review of dosing may reveal clearer associations, as doses outside of the recommended allowance may have different effects than lower ones.
Another generalization that occurs in the text is the broad use of the word “chemotherapy,” rather than focusing on the types used by patients in this study (cyclophosphamide, doxorubicin, and paclitaxel). As noted above, some supplements are, in fact, indicated to be used with particular chemotherapy drugs.
An important limitation outlined above is that the selection of antioxidants was not determined a priori but, in fact, was an unplanned secondary analysis that grouped certain antioxidants together based on a presumption of potential risk. It is unclear why “any antioxidant” was defined as vitamin A, vitamin C, vitamin E, CoQ10, and carotenoids, and this brings forth the question of which antioxidants were not included in this group and why? The large number of supplements not reported on, along with the fact that this post hoc analysis was neither predefined nor accounted for, may have contributed to bias.
Regarding omega-3 fatty acids, the authors point out that the questionnaires’ query on omega-3 fatty acid consumption lumped together all manner of different sources of omega-3s and that we cannot necessarily consider these as equivalent. In this question, omega-3s from fish were considered equal to omega-3s from dietary supplements as well as those found in flaxseed oil or cod liver oil.
Lastly and of greatest concern regarding the study limitations is the association of risk presented for each natural health product (eg, vitamin B12, iron, omega-3s) with no correction for any possible interaction between these supplements. Conceivably people taking any supplement might be more apt to take others, and as such, the effect of 1 supplement (detrimental or beneficial) might have masked or accentuated the effect of another. Without any correction for this interaction, we cannot decipher which agent might contribute more or less to the outcomes seen in this study. It is conceivable that the only harms associated may have come from use of vitamin B12, vitamin A/carotenoids, and/or iron and not from any of the other supplements at all.
Another major limitation of the study is that it did not collect information on which patients may have had prior treatment with radiation, nor patient-specific information on adherence to the cancer treatments received.
Taken together, this work by Ambrosone et al provides a good effort to address important questions and does reinforce the concept that just because something is natural does not necessarily mean it is safe; it also demonstrates that continued research is vital. From the data presented, it appears that:
- taking non-antioxidants iron and vitamin B12 during chemotherapy is also associated with worse survival; and
- that long-term use of any antioxidant (particularly vitamin A and carotenoids) that carries over into chemotherapy treatment (use before and during chemotherapy) is associated with poorer survival.
This second association is, however, tenuous and remains to be effectively addressed. To determine the true potential for harm, especially for subtle effects, we need studies with much larger populations that are based on interventional randomized trials. Observational data is good to include in the totality of evidence, especially in the absence of more rigorous data. As such, Ambrosone’s study is an important contribution to be acknowledged.
Clinicians should continue to attempt to reduce the risk of harm as much as possible while using the best evidence and clinical experience to guide patients in their care. In all, as there is limited data for benefit and the presence of some data for harm, it appears best to avoid the use of vitamin A, carotenoids, vitamin B12, and iron during the types of chemotherapy used in this study, especially in patients who were already supplementing before treatment, until more safety data is released. Overall clinician-guided supplement use vs patient-selected (or self-selected) supplement use is recommended. As for other antioxidants and supplements, it is imperative that clinicians thoroughly review the evidence on a case-by-case basis, seeking the best available data for the specific supplement and specific chemotherapy being used. Future research should focus on the timing and dosing of different supplements and their effects on outcomes, as the data presented in this study indicate that this may be an important variable.