Study design and participants

This dose-response trial utilized simple randomization to assign eligible participants at a ratio of 1:1:1 to the following groups: (1) control group; (2) low-dose group; (3) high-dose group. A random sequence of numbers (1 or 2 or 3, each corresponding to one group) was generated by the first and last authors, using Microsoft Excel; participants were enrolled following this random sequence, based on the time they joined the study. There was no blinding as participants had to consume provided nori (impossible to mask). The study advertised for participant recruitment through social media, including Facebook post of Taiwan Vegetarian Nutrition Society and Campus-wide email system and posters at Fu Jen Catholic University. Thirty vegetarians were recruited in November 2022. The trial was conducted from November to December 2022. Participants were asked to complete a demographic questionnaire (including age, sex, education, exercise habits, occupation, medical history, use of alcohol, cigarettes, and betel nuts) at enrollment.

The inclusion criteria were: age 20 to 60 years old, being vegetarian (vegan, ovo-vegetarian, lacto-vegetarian, lacto-ovo vegetarian) for at least 1 year, did not use supplements containing vitamin B12 or folate or fortified nutritional yeast in the past year. Participants were excluded if they reported having anemia, gastrointestinal diseases and surgeries, taking antacids or metformin in the past week or were alcoholic and not willing to abstain from alcoholic beverages. The study was conducted at Fu Jen Catholic University and the protocol was approved by the Institutional Review Board (approval number: C110211) at this university. All participants signed informed consent before joining the study. The trial was pre-registered at ClinicalTrials.gov (NCT05614960).

Intervention

The intervention lasted for four weeks. Control group received no nori and were instructed not to start the habit of consuming large amount of nori or any fortified foods and supplements. The intervention groups were provided with nori (for four weeks) on the day of first blood draw. Each package contained 26 sheets of nori. Low-dose group participants were provided with 5 packages and instructed to consume 4 sheets per day (5 g, initially estimated to contain 2.4 µg vitamin B12, corresponding to RDA in Taiwan and United States). High-dose group participants were provided with 8 packages and instructed to consume and 7 sheets per day (8 g, initially estimated to contain 4 µg vitamin B12, corresponding to AI in EFSA in European countries). Participants were instructed not to share this nori with family and friends, to maintain their usual diet, lifestyle and physical activities, and to continue avoiding any supplemental form of vitamin B12 (including multi-vitamins minerals) and nutritional yeast during the study period. A reminder of these study rules was printed on a post card (for participants) and on each package of nori provided. Participants who previously consume eggs, dairy, and fortified plant-milk were asked not to change the habits. The LINE smart phone application (a social media similar to What’s App, and widely used in Taiwan) had been used to regularly remind the participants to consume their prescribed daily nori.

Assessment of vitamin B12 content in nori

Vitamin B12 content of the nori samples were analyzed at the United Graduate School of Agricultural Sciences, Tottori University (Tottori, Japan). Nori samples were first extracted by KCN-boiling method, and vitamin B12 compounds were eluted and purified from extracts using Sep-Pak C18 cartridge and B12 immunoaffinity column. The purified compounds were analyzed using reversed-phase HPLC, as previously described [15].

During the pre-planning stage, four brands of commercially available unflavored roasted nori, and one fresh nori harvested from Penghu island near Taiwan were purchased and tested for vitamin B12 content, and all of them contained true vitamin B12 (rather than analogues), as detailed in Table S1. We chose one commercial brand with opaque packaging and containing the highest vitamin B12 (48.4 µg/100 g) for the intervention, as transparent packaging may expose nori to light and contribute to vitamin B12 photo-degradation.

However, when we sampled the nori from the batch used for the actual trial, it contained a lower amount of vitamin B12 (38.6 µg/100 g). This value would actually change the estimated vitamin B12 content to 1.9 and 3.1 µg for 5 g and 8 g of nori, respectively. These values were used for actual computation when assessing vitamin B12 intakes from nori. We also sampled the same brand of nori at different time throughout the year (Table S2).

Assessment of vitamin B12 nutritional status

The primary outcome of this trial was changes in vitamin B12 nutritional status (serum vitamin B12, holoTC, MMA, Hcy, and a combined score of these four markers) over the four-week intervention. Overnight fasting venous blood were collected (refrigerated at 4oC immediately) and sent to Chung-Yi Clinical Laboratory (New Taipei City, Taiwan) to assess serum vitamin B12, folate, and Hcy within the same day. Serum vitamin B12 and folate concentrations were analyzed using electrochemiluminescence immunoassay (Roche cobas e601). Hcy was analyzed using Chemiluminescent microparticle immuno assay (Abbott ARCHITECT 1L71/ABRL004/R4). The remaining blood samples (for MMA and holoTC) were centrifuged at 3000 rpm for 15 min at 4℃ shortly after collection, and stored at a -80℃ freezer for analysis of serum MMA and holoTC at the end of the trial. Serum MMA was analyzed by the Department of Chemistry, Fu Jen Catholic University (New Taipei City, Taiwan) using liquid chromatography with tandem mass spectrometry (LC–MS-MS) (Sigma-Aldrich M54058). Serum holoTC was analyzed using an ELISA kit (IBL-International) by Yi-Her Laboratory (Yilan, Taiwan). The Four combined index of vitamin B12 score (4cB12) score was calculated according to published Eqs. [16, 17] using serum vitamin B12 concentration, holoTC, Hcy, MMA, and age, as shown below:

$$\:4cB12=_\frac_}-\frac\right)}^}$$

As improvement in vitamin B12 nutritional status would result in increases in serum vitamin B12 and holoTC, and decreases in functional biomarkers including both MMA and Hcy (raise during deficiency), the equation of 4cB12 score is a combined evaluation of all four biomarkers, and its increase indicates improvement in vitamin B12 nutritional status. A 4cB12 score < -1.5 indicates possible vitamin B12 deficiency [18]. The diagnosis cut points of vitamin B12 deficiency used in this study were serum vitamin B12 concentration < 148 pmol/L [19,20,21], holoTC < 35 pmol/L [22, 23], Hcy > 12 µmol/L [24,25,26], MMA > 271 nmol/L [12, 13], 4cB12 score < -1.5 [18].

Assessment of adherence and dietary vitamin B12 intakes

A quantitative food frequency questionnaire (FFQ) – that inquires both frequency and portion size – designed specifically to assess vitamin B12 intake – had been administered (through face-to-face interview) twice, once at baseline, and once at the end of the 4-week intervention. The FFQ includes use of supplements and nutritional yeast (in the past one year); and in the past one month, consumption of 15 main sources of vitamin B12 for vegetarians and vegans in Taiwan: milk, liquid yogurt, yogurt, cheese, eggs, nori, mushroom, kimchi, fermented tofu, and the six available brands of plant-milk that are fortified with vitamin B12. If participants reported use of any supplement that could potentially contain vitamin B12 (such as multi-vitamin or B-complex supplements), they were asked to provide photos of the brand and nutrition label of the supplement to ensure that it does not contain vitamin B12. Those who consume supplemental vitamin B12 in any form were excluded from the study.

The adherence score was calculated by total intake of nori from post-intervention FFQ, divided by the study-prescribed amount of nori (according to study group assignment) x 100%. Complete adherence (consuming all nori assigned by the study) would result in a score of 100% (maximum attainable), while consuming half of the nori would result in a score of 50%.

Statistical analysis

Statistical analysis was performed using SAS version 9.4 Software (SAS Institute, Cary, NC, USA). Intention-to-treat (ITT) approach was used to analyze data. Continuous variables of baseline characteristics were presented as mean ± SD or median (p25, p75), and categorical variables were expressed as frequencies and percentages. Comparison of baseline characteristics among three groups were performed using the ANOVA or Kruskal-Wallis test (continuous variables) and Fisher’s exact test (categorical variables). The within- and among-group changes in medians of dietary vitamin B12 intakes and biochemical parameters were compared using the Wilcoxon-signed rank test and Kruskal-Wallis test, respectively. P-values below 0.05 were considered statistically significant. Post-hoc comparisons between groups were conducted using Dwas, Steel, Critchlow-Fligner procedure when Kruskal-Wallis tests were significant. General linear model was used to compare least square means of changes in each biomarker of vitamin B12, while adjusting for respective baseline value (for example, comparing changes of holoTC in three groups as outcome, while adjusting for baseline holoTC). Log transformation was applied to improve normality when the residuals of linear regression showed deviation from normality as per graphical inspection.

G*Power version 3.1.9.4 was used to perform sample size estimation. For fixed-effect one-way ANOVA, alpha error = 0.05, power = 0.80, a sample size of 24 participants were needed to detect an effect size of 0.74 – estimated based on our preliminary data of another study (in which vegetarian participants adopting a vegan diet were instructed to consume 5 g of nori per day). We therefore planned for 30 participants (10 in each group), allowing potential drop out or non-adherence. We also conducted post-hoc power analysis for each of the vitamin B12 biomarkers using the effect size derived from the current study.

Graphical plots (violin plots and bar charts) were made using Matplotlib (version 3.8.3) in Python (version 3.12.2) [27].