Plant-Based Diets and Cancer Prognosis: a Review of Recent Research

Purpose of Review Although plant-based diets are recommended for cancer prevention, their role in cancer survival is still uncertain. The purpose of this systematic review is to summarize the association between postdiagnosis plant-based diets and prognosis in cancer survivors. Recent Findings There is indication that higher intake of plant-based foods was associated with improved prognosis in cancer survivors. For colorectal cancer survival, a better prognosis was observed for a high intake of whole grains and fibre. For breast cancer survival, a higher intake of fruit, vegetable and fibre and a moderate intake of soy/isoflavone were associated with beneficial outcomes. A higher vegetable fat intake was related to improved prognosis in prostate cancer survivors. Summary Emerging evidence suggests benefits of postdiagnosis plant-based diets on prognosis in cancer survivors. However, given the high heterogeneity between studies, further research in cancer survivors, considering clinical factors (e.g. treatment, stage) and methodological aspects (e.g. timing of dietary assessment), is needed.


Introduction
Due to rising cancer incidence worldwide combined with improved survival rates especially in high-income countries as a result of advances in treatment and early detection in the past decades, the number of people living with or after cancer is growing [1]. According to current estimates, there were 50.5 million people living with cancer in 2020 who had been diagnosed within the last 5 years, and this number is projected to further increase in the next years [2]. A cancer diagnosis is recognized as a "teachable moment" [3] and is frequently observed to trigger lifestyle changes, including changes in dietary habits, in hope of improving cancer prognosis and overall health [4][5][6]. To date, however, there are no specific recommendations, and cancer prevention recommendations for the general population are applied to cancer survivors, although they may represent a specific target group with different nutritional needs and metabolic functions [7,8]. Plant-based diets are an integral part of evidence-based recommendations for primary prevention of cancer and other non-communicable diseases (NCDs), promoting a diet rich in whole grains, vegetables, fruit, nuts and legumes and a limited consumption of red and processed meat [9]. As cancer survivors are at elevated risk of premature death mainly due to their primary cancer, but also due to second primary cancers and other comorbidities, such as coronary heart disease, obesity and diabetes mellitus, these guidelines might be even more important to them than for people without a history of cancer [10]. Recent studies have investigated associations between dietary factors and survival or prognosis of various cancers, and findings indicated beneficial outcomes for higher intake of plant-based diets and components in cancer survivors [ 15•, 16]. In this context, it is important to consider several aspects, including the timing of dietary assessment (e.g. assessment after cancer diagnosis = postdiagnostic diet), different cancer sites (e.g. breast, colorectal, prostate cancer) and specific prognostic outcomes, such as cancerspecific mortality, recurrence or overall survival. To provide an overview on this topic, we conducted a systematic review of recently published meta-analyses and prospective studies investigating the association between postdiagnostic plantbased diets and overall and site-specific cancer prognosis.

Search Strategy and Study Selection
We conducted a systematic search of the recently published literature on the association between postdiagnosis plantbased diets and foods with cancer prognosis in individuals after a cancer diagnosis in PubMed from January 1, 2015 to November 9, 2021. We did not apply any restrictions or filters and used predefined search terms. We used MeSH terms and title/abstract text words related to plant-based dietary patterns (i.e. vegan, vegetarian, plant-based), food groups and foods (i.e. fruit, vegetables, whole grains, nuts, legumes, seeds, plant oils, vegetable products, soya, olive oil, tofu), nutrients or other bioactive compounds (i.e. plant protein, folate, carotenoids, polyphenols, isoflavones, glucosinolates, fibre) and cancer (i.e. cancer tumour, carcinoma), combined with prognostic outcomes (i.e. prognosis, overall survival, all-cause mortality, cancer-specific mortality/survival, recurrence), related to the study population (i.e. cancer survivors) and the study design (i.e. prospective cohort studies, systematic reviews, meta-analyses). We excluded cross-sectional and retrospective case-control studies as well as studies on prediagnosis diet. Screening of studies was conducted by at least two independent researchers. The reference lists of selected articles were additionally hand-searched for relevant literature.

Search Results
In total, 30 studies (seven meta-analyses and 23 primary studies) met our inclusion criteria (Fig. 1). Five of the 23 recently published primary studies were already considered in the identified meta-analyses, and the remaining 18 primary studies were additionally identified through the systematic literature search (Table 1). Of the seven metaanalyses [7,17,18,19 [19••] and one on overall cancer [7]. The meta-analysis on overall cancer included prospective cohort studies (Table 1).

Identification of studies via PubMed
Screening Fig. 1 Adapted flow-chart of the study selection process according to PRISMA 2020 statement [44]

Association Between Plant-Based Diets and Overall Cancer Prognosis
One meta-analysis on intake of fruit (n = 3) and vegetables (n = 4) regarding all-cause mortality in survivors of different cancer sites was identified [7]. No clear associations were observed (Table 1).

Association Between Plant-Based Diets and Breast Cancer Prognosis
For BC prognosis, we identified meta-analyses on fruit and vegetable, carotenoid and fibre intake regarding overall survival (Table 1). There was no indication for an association between fruit and vegetable intake with overall survival of BC (summary hazards ratio and 95% confidence interval (SHR (95% CI)) for high vs. low intake of fruit and vegetable combined: 0.95 (0.73, 1.24); I 2 = 17%, n = 2)). Similar associations were observed for overall survival when investigating fruit and vegetable separately (SHR (95% CI) for high fruit intake: 1.04; (0.77, 1.42); I 2 = 41%; n = 3; and for high vegetable intake: 1.08 (0.75, 1.55); I 2 = 60%; n = 3). A meta-analysis on cruciferous vegetable intake and overall survival also did not show an association (SHR (95% CI): 1.03 (0.90, 1.17); I 2 = 0%; n = 2) [17]. In line with this observation, another meta-analysis did not find an association between carotenoid intake (a proxy for fruit and vegetable intake) and overall-or BC-specific survival [18] (Table 1) Due to their oestrogen-like effects, soy/isoflavones are of particular interest for BC prognosis and showed a tendency (but imprecisely estimated) towards better overall survival in a high vs. low meta-analysis (SHR (95% CI): 0.80; (0.62,1.04); I 2 = 24%; n = 2) [22••]. In that systematic review, single study findings indicated that higher postdiagnostic soy and isoflavone intake tended to be inversely associated with BC-specific mortality (hazard ratio and 95% confidence interval (HR (95% CI)): 0.83 (0.64, 1.07)) and a decreased relative risk of recurrence (HR (95% CI): 0.75 (0.61, 0.92)) among both US and Chinese women [40]. Regarding plant-based dietary patterns, one US prospective cohort study was identified examining the association between long-term adherence to an overall plant-based diet index (PDI), a healthful PDI (hPDI) and an unhealthful PDI (uPDI) regarding BC prognosis. While for the overall PDI no associations could be observed, the hPDI was associated with a 7% reduced relative risk of all-cause death (HR (95% CI): 0.93 (0.83,1.05)) and a 17% lower relative risk of death due to other causes than BC (HR (95% CI): 0.83 (0.71,0.98)). Conversely, the uPDI was associated with increased relative risks of all-cause death and death due to other causes than BC (HR (95% CI): 1.07 (0.96, 1.20) and 1.20 (1.02, 1.41), respectively). No associations were observed between plant-based eating patterns and BC-specific endpoints [11••].
In the Nurses' Health Studies (NHS and NHS II), women with cumulative average of postdiagnostic high fruit and vegetable intake had a decreased relative risk of all-cause mortality (HR (95% CI): 0.82 (0.71, 0.94)) ( Table 2). Results indicated that especially a higher vegetable intake was associated with lower all-cause mortality (HR (95% CI): 0.84 (0.72, 0.97)), while no association was observed for fruit intake. Subgroup analyses also pointed to lower all-cause mortality risk with higher intake of cruciferous vegetables (HR (95% CI): 0.87 (0.76, 0.99)) and vegetables high in β-carotene (HR (95% CI): 0.80 (0.70, 0.91)). Findings from the NHS study cohorts provided further evidence that BC survivors should be encouraged to maintain a high intake of fruit and vegetables after diagnosis, as a decrease in total fruit and vegetable intake, or vegetables alone, by one or more servings/day from pre-to postdiagnosis was associated with a 14% (HR (95% CI): 1.14 (1.01-1.27)) and a 16% (HR (95% CI): 1. 16 (1.02-1.30)) higher relative risk of all-cause mortality [13 •]. Other studies did not find an association between fruit and vegetable consumption or whole grain intake and BC prognosis [23,28]. A recently published study from China showed that nut consumption was associated with an up to 50% reduced relative risk of BC recurrence, metastasis or mortality. Total nut intake ≥ 17 g/ week, compared to non-consumption, was inversely associated with overall survival (OS) (HR (95% CI): 0.74 (0.52, 1.05)) and disease-free survival (DFS) (0.48 (0.31, 0.73)). Stratified analyses showed that the association was more evident among participants with a higher total energy intake for OS and among participants with early stage (I-II) BC for DFS. The associations did not vary according to nut type (i.e. peanuts, walnuts, other nuts) [30•].
No association with BC prognosis was observed for plant-based protein, whereas a modest survival benefit was reported for total protein intake [26]. Recently, it has been shown that high postdiagnostic intakes of total carbohydrates (CHO) were associated with higher relative risk of BC-specific and all-cause mortality [41]. However, a recent study suggested that the sources of CHO might have different effects on BC prognosis [24]. While higher intake of CHO from vegetables was associated with a decreased mortality risk (HR (95% CI): 0.86 (0.75, 0.97)), CHO from fruit juices, refined grains or potatoes were associated with increased relative risk (HR (95% CI) for CHO from fruit juice: 1.15 (1.01, 1.30); for CHO from refined grains: 1. 16 (1.02, 1.32); and for CHO from potatoes: 1.13 (0.99, 1.28)). Similar associations were observed for BC-specific mortality (Table 2).
Addressing the controversially discussed association between soy/isoflavone intake and BC prognosis, a recently published study indicated that a higher postdiagnostic dietary intake of isoflavone was associated with reduced all-cause mortality following a linear dose-response trend. The strongest association was observed for ≥ 1.5 compared to < 0.3 mg/d (HR (95% CI): 0.65 (0.41, 1.00)). In subgroup analyses, the association was stronger for women with ER-PR-tumours and women who did not receive hormone therapy in the past [31]. A study from Hong Kong reported favourable associations with prognostic outcomes only for moderate intake of soy isoflavone, but not for the highest intake [25 •]. This association was stronger in premenopausal women, women with triple-negative tumours and women who received initial hormone treatment [25•]. Since the mean soy isoflavone intake of this Chinese study cohort was 8.5 mg/d in the early postdiagnosis period, moderate intake here may be comparable to a high intake in the North American population (mean intake of 1.8 mg/d) [31].
Results of a biomarker study from Germany of long-term postmenopausal BC survivors did not point to a clear overall relationship between circulating phytoestrogen metabolites in the blood and prognosis. However, there were some surprising findings on individual phytoestrogen metabolites: Higher concentrations of luteolin were associated with an increased relative risk of BC-specific mortality. In addition, higher concentrations of genistein and resveratrol were associated with a higher relative risk of BC recurrence [27•]. Another biomarker study indicated that higher circulating folate concentrations were related to decreased relative risk of all-cause mortality (HR (95% CI): 0.41 (0.19, 0.90)) [29].
The importance of a healthy plant-based diet for improving overall survival was further supported by findings of the CALGB 89803/Alliance trial. For colon cancer survivors, intake of ≥ 5 servings of fruit and vegetables per day was associated with improved overall survival. In the same study, it was shown that the quality of consumed grains might play a role in colon cancer prognosis [12•, 35].
A higher consumption of refined grains (≥ 3 vs. < 1 serving/day) was associated with an 88% higher relative risk of overall death (HR (95% CI): 1.88 (1.25, 2.85)). Similar findings were observed for colon cancer recurrence and diseasefree survival (DFS). No clear associations were observed for intake of whole grain with colon cancer prognosis. However, replacing a daily serving of refined grains with whole grains was associated with a 13-14% lower relative risk of overall mortality, recurrence and DFS, respectively (HR (95% CI) Findings from the NHS and HPFS cohorts in CRC survivors [15•] indicated that higher intake of whole grains was associated with better survival (HR (95% CI) per 5 g/d increase for all-cause mortality: 0.88 (0.80, 0.97) and for CRC mortality 0.72 (0.59, 0.88)). In the same study [15•], each increase in total fibre intake of 5 g/d was associated with a 14% lower relative risk of overall death (HR (95% CI): 0.86 (0.79, 0.93)) and a 22% decreased relative risk of CRC-specific death (HR (95% CI): 0.78 (0.65, 0.93)). The association was particularly present for cereal fibre for all-cause mortality (HR (95% CI): 0.78 (0.68, 0.90)) and CRC-specific mortality (0.67 (0.50, 0.90)). In addition, a 5 g/d increase in total fibre intake from pre-to postdiagnosis was associated with decreased all-cause and CRC-specific mortality.
High consumption of nuts, especially tree nuts, was also associated with a reduced relative risk of overall death and cancer recurrence in 826 patients with stage III colon cancer. Compared to non-consumers, participants who consumed ≥ 2 servings of nuts per week had a HR (95% CI) of 0.43 (0.25, 0.74) for OS and of 0.58 (0.37, 0.92) for DFS. Subgroup analyses showed that the beneficial effects of nut intake were particularly attributable to tree nut intake [32]. However, postdiagnostic vegetable fat intake was not clearly associated with cancer recurrence or mortality in the CALGB 89803 trial (HR (95% CI) for DFS comparing high vs. low intake: 1.17 (0.84, 1.62)) [36].
The hypothesis that higher folate intake after CRC diagnosis might increase mortality risk after tumour resection was not supported in the NHS and the HPFS cohorts (HR (95% CI) for high vs. low total folate intake for overall mortality: 1.04 (0.60, 1.82) and for CRC-specific mortality: 0.87 (0.65, 1.16)) [34].
A biomarker study provided little evidence that postdiagnosis serum concentrations of flavonoid phytoestrogens, for which anticarcinogenic effects have been found in experimental studies, were associated with CRC prognosis [33]. Neither serum genistein (isoflavone) nor luteolin (flavone) were associated with overall mortality, CRC-specific mortality, CRC recurrence and/or disease-free survival ( Table 2). However, the association might be different according to adjuvant chemotherapy received.

Association Between Plant-Based Diets and Urinary Tract Cancer Prognosis
We identified one study on non-muscle invasive bladder cancer prognosis that did not clearly show a protective role for postdiagnosis fruit and vegetable consumption regarding recurrence [39]. In addition, for men diagnosed with nonmetastatic PC, there is some, albeit very limited, evidence that a plant-based diet after diagnosis could improve overall survival. A higher intake of vegetable fat was associated with a decreased relative risk of overall death in the in the Physicians' Health Study (HR (95% CI) for high vs. low intake: 0.65 (0.45, 0.93)). Additionally, replacing 10% of daily calories from animal fat with vegetable fat was associated with a relative risk reduction for all-cause mortality of 44% (HR (95% CI): 0.56 (0.38, 0.80)). Higher levels of saturated fat intake were also associated with increased risk of death from all causes. No clear association was detected for PC-specific mortality (HR (95% CI) high vs. low intake of vegetable fat 0.93 (0.41, 2.14)), probably attributable to the low number of events (56 PC-specific deaths) in the cohort [37]. In the HPFS cohort, higher postdiagnostic intake of nuts (≥ 5 servings per week vs. < 1 serving per month) was associated with a 34% lower relative risk of overall death (HR (95% CI): 0.66 (0.52, 0.83)). The estimates for PCspecific outcomes also pointed to inverse associations but were imprecisely estimated ((HR (95% CI) for lethal PC: 0.88 (0.57, 1.35) and for fatal PC: 0.62 (0.36, 1.07)) [16].
As dietary lycopene has been inversely associated with the risk of incident PC, postdiagnostic lycopene intake was evaluated in relation to PC-specific mortality in the Cancer Prevention Study II Nutrition Cohort [38]. No overall associations could be observed using a single dietary measurement. However, when lycopene intake was assessed at two different postdiagnosis time points, average intake was inversely associated with PC-specific mortality but only in men at advanced stages (HR (95% CI): 0.41 (0.17, 0.99)). This finding is probably due to chance, but it is also possible that the use of data from two consecutive FFQs allowed the detection of an association that would have been obscured by measurement error if only data from one questionnaire had been used [38].

Future Research Directions
Interpretation of the evidence for dietary factors as determinants of cancer prognosis is challenging because of the broad scope of this area of research and heterogeneity between studies. There is large heterogeneity regarding study populations, exposures, their assessment (especially the timing of dietary assessment), as well as differences in the outcomes under investigation. In the future, studies are needed that are specifically designed to investigate dietary factors and cancer prognosis. These should allow examining associations between diet and cancer prognosis, including changes in dietary behaviour from pre-to postdiagnosis, in well-described populations of cancer survivors.
The findings of this systematic review are based on observational studies that are susceptible to confounding and reverse causation. The presence of symptoms, disease or treatment effects, comorbidities and overall health status might influence cancer survivors' diet, and there is a plethora of other known and potentially unknown determinants of cancer prognosis [42]. Therefore, efforts should be made to clarify these factors as comprehensively as possible so that they can be considered in future studies. Until then, it is crucial that at least the most important predictors of cancer prognosis are considered covariates, including cancer stage, grade, received treatment and comorbidities [9]. Repeated assessments of diet at multiple time points should be conducted, as this may reduce within-person variation and better represents a cancer survivors' long-term diet [13 •]. Recommendations for dietary modification after a cancer diagnosis must be supported by valid and reliable evidence. Therefore, to analyse whether postdiagnostic diet has an independent benefit on cancer prognosis, future studies need to consider dietary intake before diagnosis or dietary changes from preto postdiagnosis.
Regarding the role of plant-based diets in cancer prognosis, future studies should focus on the investigation of plantbased dietary patterns and of specific plant-based dietary approaches that exclude meat products or animal products per se (e.g. vegetarian or vegan diets). To explore which components of a plant-based diet are particularly important for cancer prognosis, individual food groups, foods, nutrients and bioactive compounds (e.g. fruit, vegetables, whole

Conclusions
Even though the evidence for postdiagnosis plant-based diets in association with cancer prognosis is very limited and studies are very heterogeneous, some preliminary conclusions can be drawn. Study findings suggest that adhering to plant-based diets might be beneficial for overall survival of cancer survivors (Table 1). The findings emphasise the importance of considering the quality of a plant-based diet for cancer survival, as healthy vs. unhealthy plant-based dietary patterns show associations in the opposite direction with mortality in BC and CRC survivors [11••, 14••]. There is initial evidence that healthy/unprocessed plant-based foods, including whole grains, nuts, fruit and vegetables may be beneficial for cancer prognosis. A high nut intake was consistently related to better survival of CRC, BC, and PC (Table 3), and, thus, further studies are needed to confirm these results. Preferring whole grains over refined grains and a high fibre intake after diagnosis seem to improve CRC prognosis. This is consistent with strong evidence that consuming whole grains and foods containing fibre decreases the risk of incident CRC [43]. The role of soy and isoflavone intake in BC prognosis remains unclear and needs further evaluation in large prospective studies from different geographic regions evaluating the whole spectrum of soy foods. To date, there is preliminary evidence that a moderate soy/ isoflavone consumption has a favourable effect on prognosis, and there is no evidence that high soy consumption is associated with adverse prognostic outcomes in BC survivors. It is important that further studies on BC survival investigate whether the associations between dietary factors and survival vary according to hormone receptor status/tumour subtype (e.g. triple negative BC), hormone therapy and menopausal status, as this may contribute to the development of targeted dietary recommendations. Promoting vegetable fat intake, particularly in substitution of animal fat, might be beneficial for overall survival after a diagnosis of PC. The potentially beneficial role of dietary lycopene intake to improve PC survival needs further investigation.
In conclusion, this systematic review shows that plantbased diets and their components might have the potential to improve cancer prognosis, especially for breast, colorectal and prostate cancer survivors. Large well-designed cohort studies, considering important clinical factors (e.g. stage, treatment) and methodological aspects (e.g. time point of dietary assessment, dietary changes) are needed to provide more robust evidence on this topic.