Greater Prediagnosis Soy Food Consumption Associated with Improved Survival in Women with Lung Cancer 

Get Permission

Among all 444 patients, soy food intake was significantly associated with risk for total mortality (P = .04 for overall significance) on multivariable analysis.

—Gong Yang, MD, MPH and colleagues

As recently reported by Gong Yang, MD, MPH, Research Associate Professor at Vanderbilt University School of Medicine, Nashville, and colleagues in the Journal of Clinical Oncology,1 higher prediagnosis consumption of soy food was associated with improved overall survival in Chinese women with lung cancer in the Shanghai Women’s Health Study. In a prior analysis in the Shanghai Women’s Health Study, the investigators found a significant inverse association between soy intake and lung cancer risk among nonsmoking women.2

Estrogen signaling has been found to promote lung cancer progression, and various data show increased risk of death from lung cancer among patients using hormone replacement therapy and reduced risk with use of antiestrogens. Plant-derived estrogens, including the isoflavones found mainly in soy foods, appear to act as natural selective estrogen receptor modulators, and preclinical studies have indicated that soy and soy isoflavones can inhibit cancer cell transformation, induce apoptosis, and inhibit angiogenesis and metastasis.

Study Details

The current analysis involved 444 Chinese women with incident lung cancer identified in the Shanghai Women’s Health Study between 1997 and 2010. Prediagnosis soy food intake was assessed at enrollment into the study and 2 years later. Patients had a mean age of 66 years at diagnosis, and 92% were “never-smokers.” Mean daily soy intakes were 18.0 g for dry weight soy food, 8.8 g for soy protein, and 30.3 mg for isoflavones. Soy food intake was not associated with age at diagnosis, smoking, obesity, family history of lung cancer, tumor stage, treatment regimens, or time between baseline dietary assessment and disease

Of the 444 patients, 318 died during follow-up, including 301 patients in whom lung cancer was the primary cause of death. The median follow-up from time of diagnosis for censored patients was 36 months. Data on tumor stage and treatment were not available for a large proportion of patients.

On multivariable analysis of demographic and clinical characteristics among 301 patients (including 231 who had died) who had data available for both clinical and nonclinical variables, increasing age and advanced tumor stage were significantly associated with increased risk of overall mortality and receipt of chemotherapy was significantly associated with reduced risk. No significant associations were observed for education, cigarette smoking, overweight/obesity, receipt of surgery or radiation therapy, or adenocarcinoma vs nonadenocarcinoma.

Findings in Total Population

Soy food and isoflavone intakes were categorized by 10th, 30th, 50th, 70th, and 90th percentiles. Among all 444 patients, soy food intake was significantly associated with risk for total mortality (P = .04 for overall significance) on multivariable analysis adjusting for age at diagnosis, education, cigarette smoking, body mass index, menopausal status, history of lung cancer in first-degree relatives, intakes of total calories and fruit and no-soy vegetables, interval between first food intake questionnaire and lung cancer diagnosis, and use of NSAIDs and vitamin supplements.

Compared with the 50th percentile of soy food intake in dry weight (16.0 g/d), women in the 10th percentile (6.3 g/d) had a 42% increased risk of mortality (hazard ratio [HR] = 1.42, 95% confidence interval [CI] = 1.08–1.87) and those in the 30th percentile (11.5 g/d) had a 15% increased risk (HR = 1.15, 95% CI = 1.03–1.28), whereas those in the 70th percentile (21.4 g/d) had an 8% decreased risk (HR = 0.92, 95% CI = 0.85–0.99) and those in the 90th percentile had a 7% decreased risk (HR = 0.93, 95% CI = 0.75–1.14).

Isoflavone intakes were 10.2, 18.8, 26.5, 37.9, and 53.5 mg/d for the 10th through 90th percentiles. Hazard ratios for isoflavone intake percentiles were similar to those for soy intake percentiles, but the overall trend was only borderline significant (P = .06 for overall significance).

Adjustment for Tumor Stage and Treatment

A second multivariable analysis involved only the 301 patients with data on clinical characteristics and adjusted for tumor stage and receipt of chemotherapy, surgery, and radiation therapy in addition to the factors included in the first multivariable analysis. On this analysis, the strength of the significant association between soy food intake and mortality was increased (P = .004 for overall significance).

Compared with the 50th percentile, women in the 10th percentile had an 81% increased risk for mortality (HR = 1.81, 95% CI = 1.26–2.59) and those in the 30th percentile had a 25% increased risk (HR = 1.25, 95% CI = 1.09–1.42), whereas those in the 70th percentile had a 12% decreased risk (HR = 0.88, 95% CI = 0.80–0.97) and those in the 90th percentile had an 11% decreased risk (HR = 0.89, 95% CI = 0.68–1.16). Thus, risk of death decreased with increasing soy food intake until intake reached approximately the 70th percentile, with further increase in intake appearing to confer no additional benefit.

Results for isoflavone intake were similar: Hazard ratios relative to the 50th percentile were 1.76 for the 10th percentile, 1.24 for the 30th percentile, 0.87 for the 70th percentile, and 0.89 for the 90th percentile. The overall trend was significant (P = .007).

The effect of soy food intake was stronger in never-smokers, in whom hazard ratios relative to patients in the 50th percentile were 2.40 in the 10th percentile, 1.42 in the 30th percentile, 0.85 in the 70th percentile, and 0.92 in the 90th percentile (P = .002 for overall significance).

Additional analysis by soy food intake tertile in the 301 patients (1st tertile, > 20.6 g/d; 2nd, 12.4–20.6 g/d; 3rd, ≤ 12.3 g/d) showed that the adjusted 12-month survival rate was 60% for women in the highest tertile and 50% for those in the lowest tertile.

As noted by the authors, since dietary assessments occurred before cancer diagnosis, no inferences on the effect of postdiagnosis diet can be made; they cite the need for investigation of the potential effects of postdiagnosis soy food intake and lung cancer outcomes, particularly in patients with early-stage disease, in whom any postdiagnosis intervention, including diet modification, might be expected to have the greatest impact. They also note that since the study was conducted among women with a low prevalence of cigarette smoking and postmenopausal hormone use, it would be useful to study the effects of soy diet among smokers and women using hormone replacement therapy.

The investigators concluded:

[T]his…study provides the first evidence that soy food consumption before cancer diagnosis may favorably affect clinical outcomes of lung cancer in women. More epidemiologic studies are needed to confirm this finding and provide support for launching randomized trials. Given that the incidence of lung cancer among women is increasing steadily worldwide and that soy can be readily incorporated into most diets, our findings, if confirmed by future studies, could potentially contribute to the development of new strategies for control and management of this fatal malignancy. ■

Disclosure: The study authors reported no potential conflicts of interest.


1. Yang G, Shu X-O, Li H-L, et al: Prediagnosis soy food consumption and lung cancer survival in women. J Clin Oncol. March 25, 2013 (early release online).

2. Yang G, Shu XO, Chow WH, et al: Soy food intake and risk of lung cancer: Evidence from the Shanghai Women’s Health Study and a meta-analysis. Am J Epidemiol 176:846-855, 2012.