Provision of low‐aflatoxin local complementary porridge flour reduced urinary aflatoxin biomarker in children aged 6–18 months in rural Tanzania

Abstract Aflatoxins are toxic secondary metabolites of fungi that colonize staple food crops, such as maize and groundnut, frequently used in complementary feeding. In preparation for a large trial, this pilot study examined if provision of a low‐aflatoxin infant porridge flour made from local maize and groundnuts reduced the prevalence of a urinary aflatoxin biomarker in infants. Thirty‐six infants aged 6–18 months were included from four villages in Kongwa District, Tanzania. The study was conducted over 12 days with a three‐day baseline period and a 10 days where low‐AF porridge flour was provided. Porridge intake of infants was assessed using quantitative 24‐h recalls by mothers. Household food ingredients used in infant porridge preparation and urine samples were collected on Days 1–3 (baseline) and 10–12 (follow‐up). Aflatoxins were measured in household foods, and AFM1 was measured in urine. At baseline and follow‐up, 78% and 97%, respectively, of the infants consumed porridge in the previous 24 h, with a median volume of 220 mL (interquartile range [IQR]: 201, 318) and 460 mL (IQR: 430, 563), respectively (p < 0.001). All 47 samples of homemade flour/ingredients were contaminated with AFs (0.3–723 ng/g). The overall prevalence of individuals with detectable urinary AFM1 was reduced by 81%, from 15/36 (42%) at baseline to 3/36 (8%) at follow‐up (p = 0.003). Provision of low‐aflatoxin porridge flour was acceptable to caregivers and their infants and successfully reduced the prevalence of detectable urinary AFM1 in infants, thus, confirming its potential to be tested in future large‐scale health outcomes trial.

Consumption of AF-contaminated foods, especially those emanating from susceptible cereals such as maize, oily seeds, and nuts such as groundnuts, is a global public health concern. The global magnitude of exposure to these toxins can vary due to climate, market characteristics, dietary availability and preferences, and food safety control systems. Countries with diversified diets combined with firm regulations that monitor AF levels in susceptible foods are better equipped to protect humans and livestock from ingestion of significant amount of these toxins. However, in countries with limited dietary diversity, widespread production and consumption of susceptible crops, combined with weak enforcement of regulations, ingestion of AF is frequent (Bennett & Klich, 2003).
In Tanzania, maize is widely cultivated and serves as the main staple food to many households (Mtaki, 2020). In addition, it is the main component of complementary food for infants and young children (Muhimbula & Issa-Zacharia, 2010). Groundnut paste or flour is also commonly used as an ingredient in cereal-based complementary foods and a condiment in family food, both to provide flavour or in place of cooking oil where the latter is inaccessible. As such, maize and groundnuts are key ingredients of composite flour used to prepare porridge, a traditional complementary food in Tanzania and other parts of the East African region. While maize forms the cereal's major part, groundnut is added as a preblend or during cooking, providing protein, lipids and micronutrients in addition to flavor and taste (Makori et al., 2019).

Key messages
• Tanzanian infants are commonly exposed to AF through porridge consumption in the complementary feeding period.
• Interventions to test if consumption of low-AF local foods can reduce prevalence of AF biomarkers are limited.
• Provision of locally processed low-AF maize: groundnut complementary porridge flour increased porridge consumption and significantly reduced the prevalence of detectable AFM1 by 81%.
• The AFM1 biomarker appeared to be responsive to the intervention.

| Production of low-AF flours for intervention
The legal limit for processed cereals set by East Africa Community and adopted by Tanzania Bureau of Standards (TBS, 2014a, 2014b) is 10 μg/kg. To produce our low-AF flours, maize and groundnuts were sourced from Kibaigwa regional market located in Kongwa District.
Pre-screening of maize and groundnuts before purchasing was done by visual inspection for healthy and mould-free grain. Then three random subsamples of approximately 250 g each were drawn from top, middle and bottom of 100 kg bag and mixed to an aggregate sample for laboratory analysis of AF. Bags with AF at or below 5 µg/kg for maize and 20 µg/kg or below for groundnuts were purchased. Grains were visually sorted, cleaned, and further tested to achieve levels below 5 µg/kg of total AFs (Ngure et al., n.d.).
Grains were milled to obtain porridge flour containing maize and groundnuts at a ratio of 4:1 and sole groundnut flour. Both porridge (1 kg) and groundnut flour (0.5 kg) were supplied to caregivers.
Quantities were estimated to last for a week, benchmarked from the quantities reported in the baseline dietary interviews. The proportions were based on the common practice in the study area (Makori et al., 2019;Mollay et al., 2021). Always stainless steel vacuum flask (ALWAYS, China) was also provided for safe and hot storage of babys' porridge.

| Data collection and intervention
This study was designed using the Trials for Improved Practices (TIPS) methodology (Dickin and Griffiths, 1997). The study took place over 12 days, with Days 1-3 being baseline collection, Days 4-12 the intervention period, with follow-up data collection on Days 10-12.
Trained nurses from the local hospital performed all interviews and collected urine samples. Separately trained staff conducted the 24 h recalls and food collection. All interviews were conducted in Swahili.
Mother baby dyads were visited on three consecutive days at both baseline and follow-up (Table 1). On Day 1, mothers were interviewed to understand beliefs and behaviours around infant porridge feeding practices and other types of food consumed by infants. A 24-h dietary recall was conducted to guide food sampling and record the number of times the baby ate porridge the day before the visit. In this study, three types of maize-based porridge are commonly fed: (1) thin porridge that can flow off the spoon, (2) thick porridge that stays on the spoon, and (3) stiff porridge or 'ugali' dishes.
For each of the AF-prone foods reported to be consumed by the infant, a random sample of 500 g of grain/flour from household stores was collected. Where only small quantities (<1 kg) of the grain or flour was available, a sample of 100-250 g was collected. To enhance representative sampling, stocks of 20 kg or less were thoroughly mixed by inverting the storage bag about five times before sampling. Five subsamples of about 100 g each were randomly drawn and mixed to an aggregate sample. In a few households with larger grain stocks (>20 to 100 kg), a sampling spear was used to draw 10 random subsamples of about 100 g each from different parts of the storage bag to obtain an aggregate sample. In both cases, the aggregate sample was mixed thoroughly, and then a subsample of 500 g or less, depending on the stock quantity, was drawn. The samples were tightly packed in dry sealable Ziplock bags and kept at −20°C for AF analysis.
On Day 3 of the baseline study, low-AF flours were provided.
Guidance was provided to the mothers to: (a) use the provided After the final baseline visit, mothers practiced feeding porridge and other foods made from the provided flours for 1 week. One visit T A B L E 1 Schedule of visits for data collection and delivery of the pilot intervention.

Activity
Baseline (

| Analysis of AFM1 in urine
Urine samples were stored frozen at −20°C in aliquots at the laboratory located in the Kongwa District Hospital, then transported on liquid nitrogen to NM-AIST laboratory. AFM1 was analysed using the Helica Urinary AFM1 ELISA kits lot AFLM 112917 (Helica) according to the manufacturer's specification (limit of detection of 80 pg/mL) and with modifications described by Smith et al. (2017).
Urine samples were thawed at room temperature, and centrifuged using a benchtop micro-centrifuge (VWR 2400-37, rotor 2434-37) at 95g for 1 min to pellet solids. The supernatant was diluted with distilled water at a 1:10 dilution, and AFM1 was measured in duplicate according to the manufacturer's instructions. ELISA plates were read on a Microplate Reader (BIOTEK EL808) at 450 nm.
Samples measuring above the standard curve were repeated following additional dilutions. An internal sample was measured across each plate to assess inter-plate variation. Samples were analysed in duplicate and those with a coefficient of variation (CV) of greater than 7% were re-measured. Urine samples from each individual child/day were analysed separately.

| Statistical analysis
R version 4.12 was used for all analyses. Specifically, amounts of porridge consumed before and after the pilot intervention are reported in median with interquartile ranges, and compared using paired sample T-test. Wilcoxon signed-rank test was used to explain differences between frequency of consumption at baseline and follow-up visits, and compare prevalence of AFM1 biomarkers before and after the pilot intervention. Spearman's correlation and regression models were used to draw association between age, frequency and volume of porridge consumption.

| Ethical consideration
Ethical approval was obtained from Cornell Institutional Review Board (IRB)-USA, and the National Institute for Medical Research (NIMR)-Tanzania. Administrative permission to visit villages and households was sought from the National through Regional, District and village authorities. Mothers provided informed consent, using a signed written consent form in Swahili to participate in the study.

| RESULTS
Forty-two infants were assessed for eligibility, of which 36 were recruited; 6 babies were ineligible, due to illness (n = 5) and travel (n = 1) ( Figure 1). Majority of the recruited households were from two F I G U R E 1 Flow chart of the recruited infants. main ethnic groups: the Kaguru (50%) and Gogo (33%), with 17% belong to other smaller groups. About 28% of mothers attended but did not complete primary school, 63.9% completed primary school, and only 9% completed secondary school. Babies were between the age of 6-9 (n = 12), 10-12 (n = 14) and 13-16 (n = 10) months, and 52.8% were boys (Table 2). All babies had been introduced to cerealbased porridge at baseline. The majority, especially from the age of 8 or 9 months and above, were introduced to family foods in addition to porridge.
At baseline, 78% of babies fed on porridge in past 24 h; of these, 39%, 46% and 14% consumed porridge once, twice and thrice a day, respectively. In the follow-up study, 97% of babies fed on porridge in the past 24 h; of these, 11%, 26%, 54% and 9% consumed porridge once, twice, thrice and four times a day, respectively. While the proportion of babies who did not consume porridge was reduced from 8 (22%) at baseline to 1 (3%) at follow-up, the shift in porridge consumption frequency between once and four times a day also increased significantly, p < 0.001 (Figure 2). The amount of porridge consumed at baseline (median 220 mL: IQR, 201, 318 mL; range 120-509 mL) was significantly lower than that at follow-up (median 460 mL: IQR, 430, 563 mL; range 100-900 mL (p < 0.001). The increase in the volume of consumption appeared to be mostly driven by an increase in the number of daily feedings rather than an increase in any given meal (Figures 2 and 3).
This increase in porridge consumption was also witnessed by mothers during the reinforcement visit, where all mothers stated that they were happy with the provided preblended porridge and groundnut flour and that the infant seemed to like the foods made with these ingredients. Seventy-eight percent of mothers (28/36) said their infant consumed more porridge when made with the provided flour than when made as usual in the household.
In addition to thin or thick porridge, babies also consumed family foods. At baseline, stiff porridge had the highest frequency: 22.4% (37/165), mango juice (3%) and others (< 3%). In the follow-up visit, the percentage of stiff porridge consumption was reduced to 1 out of 165 observations (0.6%), with an increased consumption of fresh roasted maize (4.5%), watermelon (3.9%), orange (3.9%) and pumpkin (3.9%).  F I G U R E 2 Daily porridge consumption frequency in the past 24 h as reported by mothers at baseline and follow-up (n = 36).

| AF contamination of home-made porridge ingredients
F I G U R E 3 Total daily porridge consumption volume (mL) ranges per day at baseline (n = 28) and follow-up (n = 35), grouped by daily consumption frequency.

| Prevalence of AFM1 urinary biomarkers
Three urine samples were collected on consecutive days from each of the 36 infants at both the baseline and the follow-up visits, resulting in 216 urine samples. Out of the three samples collected from each infant, 42% (15/36) of infants had detectable urinary AFM1 in at least one sample at baseline, which was significantly reduced to 8% (3/36) (p = 0.003) at follow-up. Overall, the intervention reduced the prevalence of detectable urinary AFM1 in infants by 81%. In addition, baseline samples that were detectable (AFM1 range 80-707 pg/mL) tended to be higher than those following the intervention (range 83-143 pg/mL), and notably 6/36 at baseline exceeding 150 pg/mL of AFM1 versus none in the follow-up (Figure 4).

| DISCUSSION
This study tested if low-AF porridge flours made from local maize and groundnut altered feeding behaviour was acceptable to caregivers and their infants and how their distribution affected the urinary aflatoxin M1 biomarker in infants in preparation for a larger trial. In addition, typical infant-fed foods in this region were tested for AF.  Others (4) 4 (100%) 4.0-6.5 5.5 5.7 3 (75%) 0 At follow-up Others; rice (2) 2 (100%) 5.2-6.2 5.7 -2 (100%) -Note: Others at baseline include finger millet at 6.5 μg/kg, composite flour at 4.0 μg/kg, two rice samples at 5.7 and 5.8 μg/kg. a Limit for AFB1.
b Limit for total AFs in Tanzania.
range, groundnuts more frequently exceeding the 10 μg/kg regulation (TBS, 2014a(TBS, , 2014b. These findings are limited in number but compare similarly with other studies (Bennett & Klich, 2003;Kimanya et al., 2014;Magoha et al., 2016). The levels of AF in some samples further supports a concern for infant exposure to AFs early in life as complementary foods are introduced in many parts of the East African region (Muhimbula & Issa-Zacharia, 2010, Makori et al., 2019. Typical aflatoxin exposure can be assessed in high-risk populations using aflatoxin-albumin biomarkers (Turner and Snyder, 2021); however, their use in short-term interventions is restricted due to the long half-life of this marker. By comparison, urinary AFM1 better captures exposure changes in studies over several days/weeks. Due to the more transient nature, urinary AFM1, to some extent, reflects a combination of AF levels in food, quantities of food consumed, and a component possibly more relevant for infants, the distribution or timing of feeding activities. Thus, this study actively sought to capture 3-day measures to better assess the efficacy of the intervention. The LOD of detection in our assay was higher than some other studies (G. Ezekiel et al., 2022;Polychronaki et al., 2008), and while all sampled food at baseline was found to be contaminated, some infants that were exposed would not necessarily be detected by the urinary AFM1 measure in this study. However, the study was able to quantify those with moderate to high level of urinary AFM1, arguably those of most public health concern. In our study, urinary AFM1 was detected frequently in infants at baseline, with several samples exceeding 150 pg/mL. Following several days of using low-AF flours (<5 μg/kg), this intervention reduced both the frequency of detects and lowered the range observed of those detected at follow-up, indicating significant efficacy in reducing the burden of exposure.
Extrapolation of that association would suggest that infants in our study at baseline with levels of AFM1 above 150 pg/mL may overlap with exposures (based on AF-albumin data) of concern for infant growth. However, given the modest correlation and wide confidence intervals between the two markers reported by G. , such a comparison with our data needs to be viewed with caution.

| CONCLUSION
This study population had high frequency and quantity of consumption of porridge made with homemade AF-contaminated ingredients.
Our pilot study found high acceptability and low barriers to recommended feeding practices using the provided low-AF flours.
Building from these results, the MMT trial intervention distributed 4:1 maize to groundnut porridge flour and separate groundnut flour, using AFM1 in a cohort of infants to measure AF exposure levels and adherence to intervention.