Association of plasma vitamin C concentration to total and cause-specific mortality: a 16-year prospective study in China

Abstract

Background:

Vitamin C insufficiency occurs across many countries and has been hypothesized to increase risk of various diseases. Few prospective studies with measured circulating vitamin C have related deficiency to disease mortality, particularly in low and middle-income countries.

Methods:

We randomly selected 948 subjects (473 males and 475 females) aged 53-84 years from a Chinese cohort and measured meta-phosphoric-acid-preserved vitamin C concentrations in plasma samples collected in 1999-2000. A total of 551 deaths were accrued from sample collection through 2016, including 141 from cancer, 170 from stroke, and 174 from heart diseases. Vitamin C was analyzed using season-specific quartiles, as a continuous variable, and as a dichotomous variable based on sufficiency status (normal >28 μmol/L vs. low ⩽28 μmol/L). Hazard ratios and 95% confidence intervals were estimated using Cox proportional hazards models.

Results:

We found significant inverse associations between higher plasma vitamin C concentrations and total mortality in quartile (HR_Q4-vs-Q1=0.75, 95%CI:0.59-0.95), continuous (HR_q20umol/L=0.90, 95%CI: 0.82-0.99), and dichotomous analyses (HR_{Normal-vs-Low}=0.77, 95%CI: 0.63-0.95). We observed significant lower risks of heart disease (P_{trend-by-quantile}=0.03) and cancer deaths (P_{global-across-quantile}=0.04) for higher vitamin C, whereas the association was attenuated for stroke in adjusted models. Similar inverse associations were found when comparing normal versus low vitamin C for heart disease (HR_{Normal-vs-Low}=0.62, 95%CI: 0.42-0.89).

Conclusion:

In this long-term prospective Chinese cohort study, higher plasma vitamin C concentration was associated with lower total mortality, heart disease mortality, and cancer mortality. Our results corroborate the importance of adequate vitamin C to human health.

Background

Ascorbic acid, a water-soluble vitamin also known as vitamin C, is essential for a range of physiological functions, including the syntheses of collagen, carnitine, and neurotransmitters. Vitamin C is not physiologically synthesized in humans and must be obtained from the diet. The currently accepted definition considers serum vitamin C concentrations >28 μmol/L as normal, 11–28 μmol/L as insufficient, and <11 μmol/L as deficient.¹ Several organizations have provided recommendations for daily vegetable and fruit intake in part to maintain a healthy vitamin C intake and a good balance of other nutrients. But a large proportion of many populations do not meet the recommendations and are at risk for vitamin C insufficiency or deficiency.^2–4 Two Chinese nutrition surveys showed that 65% of the Chinese adults aged 18-65 years and 75.5% of adolescents aged 14-17 years didn’t meet the estimated average requirement (EAR) for vitamin C intake. ^3,5,6 Similarly, the National Health and Nutrition Examination Survey (NHANES) showed that the proportion of Americans who met the recommendations from American Heart Association for fruits and vegetables intake, the main sources of vitamin C, was quite low (8%) and didn’t change from 1999 to 2012.⁴

Many observational epidemiologic studies have investigated the relation of vitamin C supplementation or consumption of vitamin C-rich food to risk of different kinds of diseases in humans, but most have relied only on questionnaire data and yielded heterogeneous results.^7,8 It was possibly due to challenges in accurately assessing vitamin C intake from foods, including changes that occur with different processing, storage, and cooking methods. Circulating vitamin C concentrations integrate intake, absorption, and metabolism and may more accurately reflect typical vitamin C status - although population ranges may change across seasons due to changes in the food supply and this must be accounted for in any analysis. Very few studies have examined the association between circulating vitamin C concentration and all-cause mortality, and even fewer were conducted in the setting of a long-term prospective cohort. To our knowledge, data from three prospective cohorts showed significant inverse associations between plasma vitamin C and total mortality, as well as various causes of deaths, including cancer and cardiovascular disease.^9–13 These cohorts were conducted in Western European and American populations. People living in economically-developing areas are more likely to be vitamin C insufficient but data from prospective studies in these areas are quite limited.^14,15

We previously examined the relation of serum vitamin C levels to the incident esophageal and gastric cancer from 1999/2000 to 2007 using a case-subcohort study within the General Population Nutrition Intervention Trial (NIT) cohort in Linxian, China.¹⁵ Through 2016, the subcohort was followed for more than 16 years, accruing 549 deaths due to all causes. Long-term follow-up of this subcohort provides us the opportunity to examine associations between plasma vitamin C concentrations and different causes of deaths including cancer, cerebrovascular diseases, and heart diseases in an economically-developing country population.

Methods

Design of the NIT study and blood sample collection

The design of the Linxian NIT study and its extended follow-up have been described before.^16–18 In brief, the NIT was a randomized double-blind placebo-controlled primary intervention trial conducted in Linxian, China. A total of 29,584 eligible residents aged 40-69 years were enrolled in 1986 and nine nutrients were randomly supplemented for 5.25 years using a one-half 2⁴ fractional factorial design. In 1985, 1991, and 1996, we interviewed participants by questionnaires to collect data on basic demographic characteristics, lifestyle, and dietary intakes. In 1999 and 2000, 80% of the living participants (n=16,000) provided a fasting blood sample.

Ethics approval was obtained from the Institutional Review Board of the Cancer Hospital, Chinese Academy of Medical Sciences and the US National Cancer Institute.

Subcohort selection

From the NIT participants who provided blood samples in 1999-2000, we randomly selected 1000 subjects as the subcohort in a case-subcohort study that investigated associations between vitamin C and upper gastrointestinal (UGI) cancers.¹⁵ We later excluded 52 subjects due to inadequate MPA-preserved plasma and failure of vitamin C measurement, leaving a total of 948 members for this analysis (Supplemental Figure).

Outcome definitions

From 1991 to 2016, each month village health workers checked vital status and ascertained causes of deaths for all participants by home visits, supplemented by quarterly crosschecks of the data in the Linxian Death Registry. Through 2016, endpoint ascertainment was considered complete and loss to follow-up was minimal (n=381, or 1.3%). Causes of death were categorized into total cancer, cerebrovascular diseases, heart diseases(HD), and mortality from other causes. Cerebrovascular diseases (stroke) included both hemorrhagic and thrombotic strokes. HD included coronary, ischemic, hypertensive, and other types of heart disease. Starting from the date of blood collection, participants were censored at their last known follow-up date, date of death, or the administrative closure of follow-up for this study (March 31, 2016), whichever came first.

Laboratory analyses

Immediately after sample collection, the blood was stored on ice for 3-6 hours, separated by centrifugation, and then divided into aliquots within 24 hours. One aliquot of plasma was mixed with four parts of 6% meta-phosphoric acid (MPA) specifically to enable future analysis of plasma vitamin C. These stabilized samples were then stored at −70 Celsius degrees until they were prepared for analysis.

The methods used to measure vitamin C have been described before.¹⁵ In brief, MPA-preserved plasma samples were thawed, pipetted into aliquots, and immediately frozen for longer term storage. Samples were shipped on dry ice overnight to Craft Technologies Inc where they were stored at −70° until analysis. We used high-performance liquid chromatography for vitamin C measurements similar to the method used in NHANES¹⁹.

We included 30 randomly selected samples, including six in-house quality control (QC) samples, two pooled QCs, and 22 cohort samples each batch. Three in-house QC samples with known concentrations (4.88, 23.3, and 164.4 μmol/L) whose target values had been verified against National Institute of Standards and Technology Reference Materials were included at the beginning and end of each batch. Pooled QCs were generated from pooled serum samples from 80 cohort subjects. The overall CVs corresponding to the three in-house QC samples from the highest to the lowest concentrations were 9%, 4%, and 2%, respectively. The within-batch, between-batch, and total CVs for the pooled QC samples were 4%, 4%, and 6%, respectively. Laboratory personnel were blinded to the case status and QC samples. As previously described, we also measured H. pylori in this sub-cohort, and defined the H. pylori sero-positivity cutoff as absorbances of ≥1.0 for whole-cell antibodies. ¹⁵

Statistical analyses

We calculated the median of plasma vitamin C in the subcohort by using the sampling weights from the entire NIT cohort to represent the total population. We used Wilcoxon’s Mann-Whitney U test/K-sample test to compare the equality of medians of vitamin C concentrations by selected characteristics. Cox proportional hazard models were used to calculate hazard ratios (HRs) and 95% confidence intervals (CIs) for risk estimates of different levels of vitamin C for various causes of deaths.

Plasma vitamin C concentrations were evaluated in three ways: 1) in season-specific quartiles; 2) as a continuous variable to examine the estimated risk related to each 20 μmol/L increase (approximately half the IQR); and 3) as a dichotomous variable based on sufficiency status (normal >28 μmol/L vs. low ⩾28 μmol/L).¹ Because circulating vitamin C varied by season of blood draw, all estimates come from models adjusted for season of blood draw through stratification so that subjects were always compared only to others drawn in the same season.

Potential confounders, including age, sex, history of smoking (ever smoked regularly for six months or longer), alcohol consumption (any consumption in the past 12 months), measured body mass index (BMI) and H. pylori sero-positivity, were tabulated by categories of deaths (Table 1) and subgroups of vitamin C concentrations (Supplementary Table 1). We also collected dietary information on aggregated fruit, vegetable, and meat intake (times/ year), but adjustment for alcohol consumption, BMI, H. pylori sero-positivity, the nutrients supplementation group assignment in the NIT, fruit, vegetable, and meat intake, and month-to-month variation of vitamin C concentration did not materially change the associations between vitamin C and various endpoints (data not shown), so these variables were not considered further. The multivariate models were specifically adjusted for four sex- and age- groups (<65/65+, female/male), history of smoking (yes/no) to make a parsimonious model. Analyses were conducted using SAS version 9.3 (SAS Institute, Inc, Cary, NC).

Table 1.

Selected Characteristics of Sub-Cohort from the Linxian General Population Nutrition Intervention Trial, China, 1999-2016.

Item	Total cohort	Survivors	Death
			Total	Cancer			Stroke	Heart Disease	Other
				Sub-total	UGI	Non-UGI
n	948	399	551	141	98	44	170	174	66
Follow-up person years	10707	6291	4416	1141	789	352	1297	1439	540
Age at blood draw (y)a	64.0(7.6)	59.7(6.0)	67.0(7.3)	63.6(6.5)	64.0(7.0)	63.0(5.3)	67.1(7.2)	69.4(6.6)	67.9(7.6)
<65 (%)	53.4	78.1	35.6	56.0	53.1	61.4	35.9	20.1	31.8
≥65 (%)	46.6	21.9	64.4	44.0	46.9	38.6	64.1	79.9	68.2
Sex
Male (%)	50.1	40.0	57.0	64.5	66.3	61.4	54.7	50.6	63.6
Female (%)	49.9	60.0	43.0	35.5	33.7	38.6	45.3	49.4	36.4
Smokers (%)	32.5	25.1	37.8	45.4	44.9	47.7	35.5	32.8	40.9
Alcohol drinkers (%)	26.4	25.6	26.9	29.8	27.6	34.1	26.0	24.1	30.3
BMI (kg/m2)a	21.9(2.4)	22.1(2.2)	21.8(2.5)	21.6(2.2)	21.5(2.2)	21.9(2.0)	21.7(2.8)	21.8(2.3)	21.9(2.6)
HP positivity (%)	95.7	96.0	95.5	96.5	95.9	97.7	97.7	93.1	93.9

Abbreviation: HP, H pylori sero-positivity; UGI, upper gastrointestinal cancers including esophageal and gastric cancers;

^aValues are means (SD).

Results

Through March 2016, a total of 10,707 total person-years of follow-up accrued over a median of 13.4 years of observation. Table 1 shows baseline demographic characteristics of the subcohort. There were 551 deaths out of 948 total subjects, including 141, 170, 174, and 66 from cancer, stroke, HD, and other causes, respectively. UGI cancers were the predominate cancers in this population, accounting for 98 (69.5%) of the total cancer deaths. Accidents (n=18), unknown diagnoses (n=11), and hepatocirrhosis (n=6) were the top three causes of other deaths. Compared with the survivors, subjects who died were older, more likely to be male, and more likely to be smokers. Of note, over 95% of this sub-cohort’s members were H. pylori sero-positive.

Table 2 shows plasma vitamin C concentrations by selected characteristics. The median and IQR of plasma vitamin C for the total subcohort were 32.7 and 41.6μmol/L. Overall, female participants younger than 65 years old at the time of blood draw, and never-smokers had significantly higher median levels of vitamin C than males, older participants, and ever-smokers. Samples collected in winter (November and December 1999) had a significantly higher median level of vitamin C than those collected in spring (March and April 2000) (51.5 vs. 16.8μmol/L, P<0.0001). Accordingly, further risk analyses were adjusted for age and sex subgroups, smoking status, and season of blood draw through stratification (Table 3).

Table 2.

Plasma Vitamin C Concentrations in the Sub-cohort of the Linxian General Population Nutrition Intervention Trial by Selected Characteristics, China, 1999-2016^a

Characteristic	No. of subjects	Percentile of vitamin C concentrations (μmol/L)						P valueb
		IQR	5th	25th	50th	75th	95th
Sub-cohort	948	41.6	5.0	13.5	32.7	55.2	72.6
Season of blood drawc
Winter, 1999	457	27.5	11.4	34.8	51.5	62.3	76.4
Spring, 2000	487	23.0	3.7	9.1	16.8	32.2	64.9	<0.0001
Sex
Male	473	39.2	3.6	10.7	26.7	50.0	70.2
Female	475	41.0	6.0	16.4	37.3	57.4	75.3	<0.0001
Age at blood draw
<65 year	506	39.2	5.3	15.4	34.4	54.6	72.3
≥65 year	442	43.0	4.0	11.4	26.3	54.4	71.4	0.03
Smokingd
No	637	41.0	5.4	15.4	35.3	56.5	73.5
Yes	307	37.7	3.6	5.5	25.9	48.0	71.4	0.0003

^aPlasma Vitamin C concentration were weighted, accounting for age and sex, to the entire original General Population Nutrition Intervention Trial cohort.

^bWilcoxon’s-Mann-Whitney test/K-sample test for the equality of medians.

^cBlood samples were collected in winter (November and December) of 1999 and in spring (March and April) of 2000.

^dSmoker was defined as ever smoking cigarettes for 6 or more months.

Table 3.

HR and 95%CIs for the Association between Plasma Vitamin C Concentrations and Risk of Different Causes of Deaths in the Sub-cohort of Linxian General Population Nutrition Intervention Trial, China, 1999-2016

Causes of Deaths	Plasma vitamin C
		No. of cases	Quartilea, crude HRb				Quartilea, adjusted HRc				Continuousd
			HR	95% CI	Global Pe	Ptrendf	HR	95% CI	Global Pe	Ptrendf	HR	95% CI
All deaths	Q1	170	1.00		<0.0001	<0.0001	1.00		0.04	0.006	0.90	0.82-0.99
	Q2	139	0.79	0.63-0.99			0.91	0.73-1.14
	Q3	125	0.65	0.52-0.82			0.76	0.60-0.96
	Q4	117	0.60	0.48-0.76			0.75	0.59-0.95
Cancer	Q1	52	1.00		0.002	0.007	1.00		0.04	0.08	0.93	0.78-1.12
	Q2	29	0.54	0.34-0.85			0.63	0.40-1.00
	Q3	26	0.45	0.28-0.71			0.53	0.33-0.85
	Q4	34	0.59	0.38-0.90			0.72	0.46-1.13
UGI cancer	Q1	37	1.00		0.0004	0.009	1.00		0.04	0.08	0.87	0.70-1.08
	Q2	23	0.61	0.36-1.02			0.71	0.42-1.20
	Q3	14	0.34	0.18-0.62			0.40	0.21-0.74
	Q4	24	0.58	0.35-0.97			0.73	0.43-1.24
non-UGI Cancer	Q1	16	1.00		0.2	0.3	1.00		0.3	0.5	1.07	0.78-1.45
	Q2	6	0.36	0.14-0.92			0.42	0.16-1.08
	Q3	12	0.67	0.32-1.41			0.79	0.37-1.69
	Q4	10	0.55	0.25-1.22			0.68	0.30-1.52
Stroke	Q1	51	1.00		0.1	0.02	1.00		0.5	0.2	0.91	0.77-1.08
	Q2	42	0.81	0.54-1.22			0.93	0.62-1.41
	Q3	45	0.79	0.53-1.18			0.92	0.61-1.38
	Q4	32	0.57	0.37-0.89			0.72	0.46-1.13
Heart Diseases	Q1	50	1.00		0.03	0.003	1.00		0.1	0.03	0.83	0.71-0.98
	Q2	50	0.96	0.65-1.41			1.09	0.74-1.62
	Q3	41	0.72	0.48-1.09			0.80	0.53-1.22
	Q4	33	0.54	0.35-0.85			0.65	0.41-1.02
Other	Q1	17	1.00		0.7	0.6	1.00		0.7	0.9	0.94	0.72-1.22
	Q2	18	1.00	0.52-1.95			1.20	0.62-2.35
	Q3	13	0.68	0.33-1.41			0.82	0.39-1.70
	Q4	18	0.92	0.48-1.79			1.17	0.60-2.31

Abbreviation: UGI, upper gastrointestinal cancers including esophageal and gastric cancers.

^aPlasma Vitamin C quantile concentration (in μmol/L) of the sub-cohort by season-specific quartiles (winter/spring): Q1: <34.8/9.1; Q2: 34.8-51.5/9.1-16.8; Q3:51.5- 62.3/16.8-32.2; Q4: >62.3/32.2.

^bThese models were adjusted for season of blood draw through stratification.

^cThese models were adjusted for 4 age- / sex- subgroups, tobacco smoking, and season of blood draw through stratification.

^dContinuous, define as per 20 μmol/L increases, models were adjusted for 4 age- / sex- subgroups, tobacco smoking, and season of blood draw through stratification.

^e, Global P was derived from the 3 df test for the overall quartile Cox model.

^f, P for trend was derived from the models by assigning each person the ordinal value of the quartile.

Table 3 shows results of season-specific vitamin C quartile analyses on the risk of various causes of deaths. In crude analyses, we observed significantly lower risks in those with higher plasma vitamin C concentration for total mortality and major causes of death including cancer, stroke, and heart disease. When adjusted for age, sex, and smoking status, the inverse associations remained robust for total mortality, cancer-related and heart disease-related deaths, but not for stroke-related deaths. We found a strong monotonic inverse trend between quantiles of plasma vitamin C concentrations and total mortality (P_trend=0.006), and observed inverse associations between plasma vitamin C quantiles and heart disease deaths (P_trend=0.03) and cancer deaths (P_global=0.04). When further examined vitamin C concentration as a continuous variable, for each 20μmol/L increase, a 10% and 17% reduction was observed on risk of total death (HR=0.90, 95%CI: 0.82-0.99) and heart disease death (HR=0.83, 95%CI: 0.71-0.98), respectively. However, we did not see associations with higher plasma vitamin C concentrations for other causes of death.

Considering the total mortality of subjects with deficient and insufficient vitamin C concentrations were comparable and the limited cases in these two groups, we further investigated the associations between vitamin C and risk of death using the established cutoff of low (≤28μmol/L) and normal (>28μmol/L) plasma vitamin C concentrations (Table 4). There was no interaction between season and total mortality. Consistent with the inverse associations identified in higher quartiles of vitamin C concentrations, we found that participants with normal vitamin C levels had significantly lower risks of total death (HR=0.77, 95%CI: 0.63-0.95) and heart disease death (HR=0.62, 95%CI: 0.42-0.89) than those with low vitamin C concentrations. The season-specific analyses showed that the lower risk of total death (P=0.006) was specifically robust in samples collected in winter.

Table 4.

HR and 95%CIs for the Association between Plasma Vitamin C Status^a and Risk of Different Causes of Deaths, China, 1999-2016

Causes of Deaths	Case No.	HRb	95% CI	P value
All deaths	551	0.77	0.63-0.95	0.01
Winter, 1999	272	0.67	0.50-0.89	0.006
Spring, 2000	279	0.89	0.67-1.18	0.41
Pinteraction with season	0.2
Cancer	141	0.78	0.52-1.16	0.22
UGI cancer	98	0.68	0.42-1.11	0.12
non-UGI Cancer	44	1.09	0.53-2.24	0.81
Stroke	170	0.91	0.63-1.31	0.60
Heart diseases	174	0.62	0.42-0.89	0.01
Other	66	0.88	0.48-1.61	0.69

Abbreviation: UGI, upper gastrointestinal cancers including esophageal and gastric cancers.

^aPlasma vitamin C status, compare normal (>28umol/L) to low (≤28umol/L).

^bThese models were adjusted for 4 age- / sex- subgroups, tobacco smoking, and season of blood draw through stratification.

Discussion

Through 16.4 years of follow-up, we found significant inverse associations between plasma vitamin C concentrations and total mortality, cancer mortality and heart disease mortality. Our data showed that people with the highest plasma vitamin C quartile concentrations had a 25% lower risk of total death than those with the lowest quartile. Similarly, when compared to people with biochemically-defined low vitamin C status, participants with normal blood values had a 23% lower risk of death overall and a 38% lower risk of heart disease death.

This study is the first to find general benefits for higher plasma vitamin C concentrations on total and cause-specific mortalities, including cancer and heart diseases, in a long-term prospective cohort from China. Similar general beneficial results have been reported in studies from economically-developed Western populations. The European Prospective Investigation into Cancer and Nutrition (EPIC) study was conducted in healthy European populations and showed that plasma vitamin C levels were inversely related to risk of gastric cancer and mortality from cardiovascular disease.^9–11 The NHANES II and III in the USA measured plasma vitamin C at baseline and found a consistent apparent general benefit of higher vitamin C on total and cancer mortality in the follow-up studies, but that was not the case for cardiovascular disease.^12,13 A UK study conducted in people over 65 years of age found similar beneficial associations between higher circulating vitamin C and total mortality.²⁰ Other epidemiological studies conducted in economically-developed countries have likewise found beneficial associations with higher plasma vitamin C concentrations, but only for selected disease, such as cancer, cardiovascular disease, or stroke.^9,11,21–23

This study adds new evidence suggesting a robust overall benefit of higher plasma vitamin C concentrations on various causes of death previously observed in developed countries is also evident in an economically developing country. Together with results from other cohorts in Western countries, we see growing evidence that higher vitamin C status may be related to a lower total mortality.

Globally, a substantial proportion of populations are vitamin C deficient, but the situation in low-income populations is the worst. Using <11 μmol/L as a cutoff, the prevalence of vitamin C deficiency was around 7%-10% in the NHANES Study in US,²⁴ 20% in a study of Scotland,²⁵ and 1% in a general population of France.²⁶ A summary of micronutrients intake across 8 European countries showed that the proportion of vitamin C intake under EAR ranged from 8% to 13% in Poland, lower in UK and other countries.²⁷ In contrast, the prevalence of vitamin C deficiency was much higher in developing countries such as India where it is reported as 74% in northern India and 46% in southern India.¹⁴ In our study in rural China, the overall prevalence of vitamin C deficiency was 21%, ranging from 5% in winter to 35% in spring. This was consistent with a study conducted in a metropolitan population of China that showed a seasonal variance of blood vitamin C concentration, with the highest in winter and lower in other seasons.²⁸ The seasonal variations on vitamin C concentration might be more robust in regions of lower socioeconomic status due to a higher dependence on the seasonal foods consumption and local food availability patterns. In the NIT cohort, only 1.5% of people regularly took vitamin or mineral supplements. In Linxian, more vitamin C sources were consumed in winter than in the spring timepoint of our collection, which was too early in the growing season for local fresh food to have a substantial effect. People who were insufficient for vitamin C in winter (the peak season) are likely to be deficient year round, while subjects who were insufficient for vitamin C in spring might be classified as normal in winter. This pattern would lead to attenuation of the inverse association between being classified as having normal vitamin C and total deaths in the spring season.

There are several possible biological mechanisms that could explain the inverse associations we observed. Oxidative stress is one of the causative factors of HD and cancer.^29,30 It can lead to endothelial dysfunction through reduction in nitric oxide (NO) availability, inflammatory response, and lipid peroxidation.²⁹ Vitamin C may prevent oxidative stress and improve NO production of the endothelium, which, in turn, increases vasodilation and reduces blood pressure.³¹ In addition, vitamin C can prevent atherogenesis by scavenging reactive oxygen species (ROS) to protect from the oxidative damage of low-density lipoproteins, and interrupting inflammatory processes.³² Moreover, vitamin C can protect cells from oxidant DNA damage, inhibit formation of N-nitroso compounds, and neutralize the toxicity of ROS caused by H. pylori infection.¹⁵ Studies have also shown that vitamin C may serve as epigenetic regulators, promoting the demethylation of DNA and histones.³³

Several primary prevention trials have been conducted to test the potential effect of vitamin C supplementation on cancer or cardiovascular disease prevention. Contrary to the largely beneficial results reported for higher vitamin C status in observation studies, trials that tested vitamin C as a single agent, mostly in replete nutrient-populations, have not reported benefits (e.g., the Physicians’ Health Study II ^8,34,35 and the Women’s Antioxidant Cardiovascular Study ³⁶). In contrast, multi-agent trials conducted in nutrient-deficient populations, such as the NIT cohort, in which the current study was nested, showed a 10-year sustained protective effect on stroke for combined daily supplementation with 120 mg vitamin C and 30 μg molybdenum (HR=0.92, 95%CI=0.86-0.99), and some evidence for total mortality (HR=0.97, 95%CI=0.94-1.01).¹⁷ The Shandong Intervention Trial conducted in China also found a lower mortality of gastric or esophageal cancer for combined supplementation with vitamin C, vitamin E and selenium.³⁷ Baseline nutritional status may act as the fundamental factor for the different effects reported in different trials. It is also possible, that these inconsistent results may be explained by differences in the effects of vitamin C when it’s consumed as purified supplement or combinations as opposed to when it’s consumed in individual foods or certain dietary patterns, which contain complex mixtures of various dietary fiber, vitamins and minerals, and account for food synergy effects.^38,39 And ranking of exposures in observational studies may represent difference in long-term exposure over a greater period than those tested in clinical trials. Therefore, acquiring a more complete understanding of vitamin C in individual foods, how vitamin C and other substances in foods interact with one another, and factors that influence the uptake and distribution of food-derived vitamin C in the body should continue to be active areas of ongoing cancer prevention research.

This study has several potential limitations. All associations found in observational studies could be due to residual confounding. Assessment of vitamin C occurred at only one timepoint, with the underlying assumption that a single value is indicative of normal long-term intake and blood concentration. Although there is tremendous seasonal variation on vitamin C status, we did season-specific analyses and used season-specific quartiles which would account for the seasonal variations. Degradation of vitamin C during long-term storage may be a concern, but a previous study showed no degradation in human plasma frozen for five years at −80° following stabilization by MPA.⁴⁰ In addition, vitamin C might also be related to other types of behavior and nutrients (e.g. flavones and dietary fiber in the same food sources). Though we did not have a contemporaneous comprehensive assessment of diet in this study, we did assess the potential confounding effects of self-reported total fruits, vegetables, and meat consumption, and these variables did not materially change the results (data not shown).

In conclusion, this study provides evidence from a rural Chinese population that higher vitamin C status is associated with lower mortality, which corroborates the importance of adequate vitamin C not only in economically-developed populations but also in economically-developing regions of China.

What is already known on this subject

Vitamin C insufficiency occurs across many countries and has been hypothesized to increase risk of various diseases. Very few prospective studies with measured circulating vitamin C have related deficiency to disease mortality, and the limited evidence were typically from the Western and American populations. However, people living in low and middle-income countries are more likely to be vitamin C insufficient and data from prospective studies in these areas are quite limited.

What this study adds

In this long-term prospective Chinese cohort study, we found significant inverse associations between baseline plasma vitamin C concentrations and total mortality, cancer mortality and heart disease mortality. This study adds new evidence suggesting a robust overall benefit of high plasma vitamin C concentrations on various causes of death previously observed in developed countries is also evident in a rural population from an economically developing country. Together with results from other cohorts in Western countries, we see growing evidence that higher vitamin C status may be associated with a lower risk of total mortality and corroborate the importance of adequate vitamin C to human health.