Shindoh, Abe & Yasui 2010 — Cadmium fate during miso and soy sauce processing

Shindoh and coauthors tracked cadmium mass balance through bench-scale laboratory preparation of two traditional Japanese fermented soybean condiments — salty rice miso and koikuchi soy sauce — using two soybean lots of contrasting cadmium content (0.040 µg/g and 0.320 µg/g fresh weight). The processing fate of cadmium differed substantially between the two product types. In rice miso preparation, 83% of the total cadmium in the raw soybeans remained in the final product, regardless of starting soybean cadmium level. In koikuchi soy sauce preparation, only 43% (soybean A) to 53% (soybean B) of the soybean cadmium was estimated to transfer to the final raw soy sauce at the bench-scale 60–62% nitrogen yield, with 30–40% of soybean cadmium retained in the pressed soybean cake. The authors note that miso receives material contributions of cadmium from the rice koji ingredient as well — 69% of total cadmium in soybean-A miso was attributable to rice koji rather than to soybeans — so the rice koji input becomes important to cadmium loading whenever the soybean itself is comparatively clean. For soy sauce, the soybean lot’s processing characteristics rather than starting cadmium content appear to govern transfer to the final product, with the moromi-mash pressing step identified as the key locus of cadmium partitioning between liquid (soy sauce) and solid (cake) fractions.

Key numbers

Starting cadmium content of the two soybean lots (whole soybeans, fresh weight basis):

The paper reports Mean ± S.D. (n = 2 or 3) for every cadmium value in the per-process tables (Tables 1–4); the SD values were not machine-extractable from the scanned PDF tables and are noted here in the corresponding row as ± without a numeric value when not recoverable from body text. All cadmium values below are µg/g fresh weight (fw) for content and µg total for mass balance, exactly as reported by the source.

Table 1 — Rice miso processing using soybean A (low-Cd starting material, 0.040 µg/g fw):

* Estimated assuming that no cadmium was lost after autoclave (per source footnote).

Table 2 — Rice miso processing using soybean B (high-Cd starting material, 0.320 µg/g fw):

* Estimated assuming that no cadmium was lost after autoclave (per source footnote).

For soybean A, 31% of total cadmium in the rice miso was derived from soybeans and 69% from rice koji. For soybean B, 78% of total cadmium was derived from soybeans and 22% from rice koji. The authors note that sweet miso contains more rice koji than salty miso and therefore that the cadmium content of the rice used to make koji deserves additional attention when koji is the dominant input.

Table 3 — Koikuchi soy sauce processing using soybean A (0.040 µg/g fw):

** Estimated assuming that the ratio of hydrolyzed cadmium in soybeans and wheat of moromi mash was the same (per source footnote).

Table 4 — Koikuchi soy sauce processing using soybean B (0.320 µg/g fw):

** Estimated assuming that the ratio of hydrolyzed cadmium in soybeans and wheat of moromi mash was the same (per source footnote).

For soybean A, 37% of total cadmium in the moromi mash came from soybeans and 63% from wheat; for soybean B, 82% came from soybeans and 18% from wheat. The wheat input (55 µg total cadmium per 1000 g roasted crushed wheat in this study) is the dominant source of moromi cadmium when the soybean itself is low-cadmium, but it is dwarfed by the soybean contribution when the soybean lot is high-cadmium.

Nitrogen mass balance (the basis of the cadmium yield estimates for soy sauce):

Yield of hydrolyzed nitrogen in the raw soy sauce versus total nitrogen in the raw materials: 60% (soybean A processing) and 62% (soybean B processing). The authors note that modern commercial soy sauce manufacturing achieves nitrogen yields exceeding 90% through improved pressing, which compresses cake moisture below the 49.0–53.6% values measured here, and that higher pressing efficiency would transfer more cadmium from materials into the final soy sauce — but the per-volume cadmium concentration in the final soy sauce would not necessarily rise, because the total volume of soy sauce produced also increases with better pressing.

Soybean cake moisture (the pressed solid residue after moromi filtration):

• Soybean A cake: 53.6% moisture.
• Soybean B cake: 49.0% moisture.

Both values are described as approximately double the cake moisture achieved by modern soy sauce manufacturing facilities, reflecting the limitations of the bench-scale centrifugal filter (SYG-3800-15A, Sanyo Physical & Chemical Appliances Mfg. Co., Tokyo; mesh 15A, 6000 rpm for 10 min) compared with industrial pressing.

Process-step cadmium loss in miso (soaking + autoclaving):

• Soybean A: raw 0.047 µg/g dry matter → autoclaved 0.045 µg/g dry matter (small elution).
• Soybean B: raw 0.363 µg/g dry matter → autoclaved 0.337 µg/g dry matter (small elution).
• Overall retention through soaking + autoclaving: 83% of total cadmium in raw soybeans remained in the autoclaved samples, for both soybean A and soybean B.

The remaining 17% is presumed to have eluted into the soaking and autoclaving liquor; the authors do not separately measure that liquor.

Final-product mass yields:

• Salty rice miso: ~3900 g from 2200 g autoclaved soybeans + 920 g rice koji + 440 g NaCl + water.
• Koikuchi soy sauce (raw, after pressing): 3257 g from 2200 g soybean-A autoclaved soybeans + 1000 g roasted crushed wheat + 0.4 g tane-koji + 2.7 L of 23% NaCl solution.
• Koikuchi soy sauce (raw, after pressing): 3557 g from the same soybean-B-based recipe.

Methods (brief)

Sample preparation for cadmium analysis. Each sample (soybeans, autoclaved soybeans, rice koji, roasted crushed wheat, miso, soy sauce, soybean cake) was decomposed with nitric acid and sulfuric acid in a Kjeldahl flask. Cadmium was extracted from the resulting acid solution using ammonium pyrrolidine dithiocarbamate (APDC) as a chelating reagent with butyl acetate as the organic extractant, concentrating cadmium and removing alkali metals (relevant because miso and soy sauce are high-salt matrices). The amount of chelated cadmium in the butyl acetate phase was quantified by air–acetylene atomic absorption spectrometry (AAS) at 228.8 nm using a Nippon Jarrell-Ash AA-890 spectrometer (Kyoto). A calibration curve was prepared using extracted cadmium from standard solutions. Stock standards and reagents were of the grade specified by JIS K0121 (Japanese Standard Association 2006, General rules for atomic absorption spectrometry).

The analytical procedure follows the method used by the Ministry of Agriculture, Forestry and Fisheries of Japan (MAFF) for cadmium surveys of agricultural products (MAFF reference cited in the source as http://www.maff.go.jp/j/syouan/nouan/kome/k_cd/cyosa/pdf/an2.pdf, accessed 5 Jan 2010 by the authors). Replication was n=2 or 3 per cadmium measurement (mean ± S.D. as reported in Tables 1–4).

Soybean processing. 1 kg of raw soybeans was soaked in 3 L of water overnight, drained, and autoclaved under steam pressure of 0.7 kg/cm² (112 °C) for 30 min, yielding approximately 2.20 kg of autoclaved soybeans (the moisture-driven mass increase).

Salty rice miso preparation. Target composition: 12% salt, 45% moisture. The 2.20 kg of autoclaved soybeans was mixed with 0.92 kg of rice koji (soybean : rice = 10 : 8, w/w), 0.44 kg of sodium chloride, and a quantity of water adjusted to the moisture content of each mixed material. The mixture was minced with a hand-turned meat grinder, placed into a 3-L beaker under a 0.8-kg weight for pressing, and fermented at 30 °C for 40 days.

Koikuchi soy sauce preparation. The 2.20 kg of autoclaved soybeans was mixed with 1.00 kg of roasted and crushed wheat, inoculated with 0.4 g of tane-koji (mold spore inoculum, Nihon Jozo Kogyo Corp., Tokyo), and incubated at 25–30 °C under 95% relative humidity for 3 days to make soy sauce koji. The koji was mixed with 2.7 L of 23% sodium chloride solution in a 5-L beaker to form the moromi mash, which was fermented under the following temperature programme (Shimura 1998, Nippon Jozo Kyokai Shi 93, 120–128): 15 °C for the first 20 days, 25 °C for 10 days, 30 °C for 30 days, and 25 °C for the final 90 days. Moromi was stirred occasionally during fermentation. After fermentation, the mash was filtered through a small centrifugal filter (SYG-3800-15A, Sanyo Physical & Chemical Appliances Mfg. Co., Tokyo, mesh 15A, 6000 rpm for 10 min) to separate raw soy sauce from cake.

Moisture and nitrogen. Moisture content of raw soybeans, autoclaved soybeans, rice koji, and roasted crushed wheat was determined by drying to constant weight at 130 °C (2 h for soybeans; constant-weight drying at 135 °C for rice koji and roasted wheat). Moisture content of miso, soy sauce, and soybean cake was determined after mixing with silica sand to constant weight at 70 °C under reduced pressure. Nitrogen content was determined by the improved Kjeldahl method (Yasui et al. 1977, Report of National Food Research Institute 32, 108–114). Yield of nitrogen was calculated as the ratio of total nitrogen in the raw soy sauce to total nitrogen in the raw materials.

Cadmium-contamination control. The authors state that cadmium contamination from materials other than soybean, rice koji, and wheat was considered negligible because the study used guaranteed-reagent-grade NaCl, high-purity ion-exchanged water, and borosilicate glassware. They further assume that no cadmium leaves the fermentation system except as a trace component of volatile compounds and water lost during fermentation, so that all cadmium in the autoclaved soybean + koji is presumed to remain in the final miso product; the rice miso mass-balance check (sum of cadmium in autoclaved soybean and rice koji equalling the cadmium detected in the final miso) is reported as consistent with that assumption.

Evidence Fitness

This is a bench-scale processing-fate experiment, not an occurrence survey. The peer-reviewed quantitative findings — 83% retention of soybean cadmium in rice miso and 43–53% transfer to raw koikuchi soy sauce at a 60–62% nitrogen yield — are the load-bearing results for understanding how raw-material cadmium translates into final-product cadmium for two traditional Japanese fermented soybean condiments. The single-laboratory, two-soybean-lot design is the primary tier-B limitation.

Best uses:

• Process-fate evidence that cadmium in raw soybeans is largely retained (~83%) through miso preparation, with cumulative losses concentrated in the soaking + autoclaving step (~17% elution) rather than during fermentation.
• Process-fate evidence that cadmium in raw soybeans partitions roughly 43–53% to raw soy sauce and 30–40% to pressed cake during koikuchi soy sauce preparation at the bench-scale 60–62% nitrogen yield, with the moromi-pressing step identified as the key partitioning locus.
• Documentation that the rice koji input is a material — sometimes dominant — source of cadmium in miso when the soybean lot itself is low in cadmium (69% of miso cadmium derived from rice koji when the starting soybean carried 0.040 µg/g cadmium).
• Documentation that the roasted wheat input is a material — sometimes dominant — source of cadmium in soy sauce moromi when the soybean is low in cadmium (63% of moromi cadmium derived from wheat in the soybean-A run).
• Methods reference for APDC-butyl-acetate extraction with air–acetylene AAS quantification at 228.8 nm in high-salt fermented matrices, anchored to the MAFF cadmium-survey protocol.
• Scaling caveat for any extrapolation of bench-scale yields to commercial soy sauce: modern commercial pressing achieves >90% nitrogen yield versus the 60–62% bench-scale yield here, so cadmium transfer to commercial raw soy sauce may be proportionally higher (though final concentration is buffered by the proportional increase in soy sauce volume).

Not appropriate for:

• Setting a percentile-based HMTc occurrence value for either miso or soy sauce (n=2 soybean lots, single laboratory, bench-scale preparation; this is not an occurrence dataset).
• Brand-level or commercial-product claims (the source does not characterise commercial product and the wiki does not propagate brand-level values per Part 12).
• Inferring cadmium concentrations in commercial miso or soy sauce directly — the bench preparation differs from industrial process in pressing efficiency and possibly other steps; commercial occurrence requires market-survey evidence.
• Generalising the rice-koji and wheat cadmium contributions to all rice and wheat globally — the rice koji and wheat used in this study were single purchased lots, and the koji rice cadmium of ~0.07 µg/g (back-calculated from the reported 67 µg total Cd in 920 g rice koji) is a single-source value, not a population estimate.
• Drawing inferences about cadmium speciation in fermented soybean products — the AAS method quantifies total cadmium only.

Implications

• Processing does not strongly attenuate cadmium in miso. ~83% of cadmium in raw soybeans is retained through full miso preparation, with the soaking + autoclaving step accounting for essentially the entire process-related elution (~17%). The fermentation step itself does not remove cadmium. The implication for cadmium control in miso is upstream-sourcing: control the cadmium content of the soybean and the rice (for koji), because the process does not provide downstream attenuation.
• Soy sauce processing partitions cadmium between liquid and solid fractions, with bench-scale yields giving 43–53% to soy sauce and 30–40% to cake. The moromi-pressing step is identified as the key partitioning locus. The implication for cadmium control in soy sauce is two-pronged: (i) upstream control of raw-material cadmium, and (ii) recognition that the cake fraction is a meaningful sink for cadmium that does not enter the final consumer product — though that sink is reduced as commercial pressing efficiency improves.
• The koji and wheat inputs are non-trivial cadmium contributors to fermented soybean condiments. When the soybean lot itself is low in cadmium, the rice koji (in miso) and the roasted wheat (in soy sauce moromi) can contribute the majority of cadmium to the final product. Any future synthesis of cadmium in fermented soybean condiments needs to account for koji-rice and wheat cadmium independently, not just the headline soybean value.
• Bench-to-commercial scaling of soy sauce yields is non-trivial for cadmium mass balance. Modern commercial nitrogen yields (>90%) exceed the bench-scale 60–62% reported here, meaning more raw-material cadmium would transfer to the final commercial soy sauce. The authors note that the final per-volume concentration in commercial soy sauce need not be higher than in the bench-scale soy sauce — because the total volume of soy sauce also increases — but the per-mass-of-soybean cadmium yield to the final product would be higher.
• The Enrei soybean variety used as the low-cadmium reference lot here (0.040 µg/g fw) is a commonly cultivated Japanese variety; the high-cadmium contrast lot (0.320 µg/g fw) is at the high end of typical soybean cadmium values reported in the Japanese cereal-and-pulse cadmium literature (Ikebe et al. 1991, Shokuhin Eiseigaku Zasshi 32, 48–56). The bench experiment therefore brackets a realistic range of starting soybean cadmium for Japanese-style miso and soy sauce production.

Provenance notes

• Publication: Report of the National Food Research Institute (食総研報, Rep. Nat’l Food Res. Inst.), No. 74, technical-report section (技術報告), published 2010. The PDF carries Japanese date marks “平成 21年10月30日受付 / 平成 21年10月30日受理” indicating received and accepted on 2009-10-30. No DOI is assigned; institutional reports of this kind are not routinely DOI-minted.
• Corresponding author: Kumiko Shindoh, National Food Research Institute, National Agriculture and Food Research Organization (NARO), 2-1-12 Kannondai, Tsukuba, Ibaraki, 305-8642 Japan; email shindoh@affrc.go.jp (per the paper’s masthead).
• Co-authors: Takashi Abe (Japan Food Research Laboratories, 7-4-41, Saito Asagi, Ibaraki, Osaka, 567-0085 Japan) and Akemi Yasui (NARO, same address as Shindoh).
• Acknowledgments name the late Dr. S. Nikkuni for advice and identify funding as part of the Endocrine Disrupters Project of the Ministry of Agriculture, Forestry and Fisheries of Japan.
• License classification: the Report of the National Food Research Institute is a publicly-distributed institutional report of NARO, a Japanese national research organisation. Treated here as open-access-government-report-japan. Distribution terms beyond the implicit public-domain-of-government-research norms in Japan are not explicitly stated on the PDF.
• The source PDF in this corpus was sourced via the Kimi Agent Download Corruption Issue manual-fetch batch (folder condiments2_papers/03_Condiments/, file 05_Shindoh_cadmium_soy_sauce_processing.pdf).

Verification notes

• Brand-firewall posture (Part 12, strict). The source paper does not name any commercial brand of miso or soy sauce. The two soybean lots are coded “soybean A” and “soybean B” throughout; soybean A is identified by variety name (Enrei), and soybean B is identified only as a different variety of higher cadmium content. The materials section identifies reagent and equipment vendors (Nihon Jozo Kogyo Corp. for tane-koji, Kanto Chemical for NaCl, Millipore for water, Sanyo Physical & Chemical Appliances for the centrifugal filter, Nippon Jarrell-Ash for the AAS) which are retained per Part 12 Exception 2 (scientific-method vendor/material names). No brand redaction was required.
• Methods-vendor names (Part 12 Exception 2). Vendor names are retained where they identify the instrument, reagent, or material: Nippon Jarrell-Ash AA-890 AAS, Sanyo Physical & Chemical Appliances Mfg. Co. SYG-3800-15A centrifugal filter, Millipore Super Q ion-exchange water system, Nihon Jozo Kogyo Corp. tane-koji, Kanto Chemical Co. guaranteed-reagent-grade NaCl. These are scientific-reproducibility identifiers, not contamination rankings, and do not violate Part 12.
• Speciation: total cadmium only. AAS at 228.8 nm with APDC-butyl-acetate extraction is a total-cadmium method. Cadmium has no speciation convention in HMI taxonomy (unlike arsenic or mercury), so metals: [Cd] is the complete metal frontmatter for this source.
• Ingredient routing. ingredients: ["[[ingredients/soy]]", "[[ingredients/rice]]", "[[ingredients/wheat]]"] uses three existing taxonomy slugs corresponding to the three raw-material ingredients tracked by the cadmium mass balance: soybeans (soy), rice (as rice koji), and wheat (as roasted crushed wheat). The wiki has no specific soybean slug separate from soy, no koji ingredient slug, and no rice-koji ingredient slug; rice is the upstream ingredient and koji is a processing form of it.
• Product routing. products: ["[[products/condiments-general]]", "[[products/soy-products]]"] uses two existing taxonomy slugs. The condiments-general slug is the routing destination for soy sauce (consistent with the precedent established by rodriguez2009-fish-sauce-arsenic-speciation.md for fish sauce and bergerritchie2013-hot-sauce-lead.md for hot sauce, both fermented/processed condiment sources in this same Kimi Agent Download Corruption Issue folder). The soy-products slug is the routing destination for miso and other fermented soy foods. There is no dedicated soy-sauce, miso, or fermented-soy-condiment product slug in the current taxonomy snapshot (generated 2026-05-18); the gap is surfaced for Karen’s taxonomy review.
• Matrix vocabulary. matrices: [soy-sauce, miso, soybeans, soybean-paste, fermented-soybean, rice-koji, moromi, koikuchi-soy-sauce, salty-rice-miso] captures the as-tested matrices (soy sauce, miso) at both the umbrella level (soy-sauce, miso) and the as-prepared specificity (koikuchi-soy-sauce, salty-rice-miso) so future routing can fan to either level. soybeans is the raw-material matrix; soybean-paste is the English-language process-category for miso; fermented-soybean is the broader process category that links miso with other fermented soy foods; rice-koji and moromi are the in-process intermediates that the cadmium mass balance tracks. Matrices vocabulary is bare strings, not wikilinks; soy-sauce and soybeans are pre-existing values in the matrices snapshot, the others are introduced here as descriptive matrix labels.
• Jurisdictions. jurisdictions: [JP] reflects Japan as both the country of origin of the soybean varieties studied and the country of laboratory preparation (Tsukuba, Ibaraki). The product categories (koikuchi soy sauce, salty rice miso) are specifically Japanese-style fermented soybean condiments.
• Evidence-tier B rationale. Peer-reviewed institutional technical report with replicated cadmium analyses (n=2 or 3 per sample) and explicit mass-balance accounting. Methodological strength: APDC-butyl-acetate extraction is the published MAFF cadmium method for high-salt matrices, and the AAS detection limit is well below the measured cadmium concentrations. Primary tier-B limitations: n=2 soybean lots, single-laboratory bench-scale preparation rather than commercial process, and the scanned-PDF tables (Tables 1–4) carry numerical entries (moisture %, Cd content µg/g fw, weight g, total Cd µg, yield %) that did not extract cleanly through the OCR pipeline. The tabular standard deviations therefore appear as ± placeholders in the Key numbers tables above; the numerical values were verified against the body-text discussion of the same tables, which independently states the 83% miso retention, 43–53% soy sauce transfer, the 31%/69% and 78%/22% miso source-attribution splits, the 37%/63% and 82%/18% moromi source-attribution splits, the 60% and 62% nitrogen yields, the 53.6% and 49.0% cake moisture values, and the 3257 g and 3557 g soy sauce yields.
• Year attribution. The PDF carries 平成 21年10月30日受付 / 平成 21年10月30日受理 (received and accepted 2009-10-30); the journal is Rep. Nat’l Food Res. Inst. No. 74, which was published in 2010 (the references include a MAFF URL “Accessed 5 Jan 2010” indicating editorial-period dating consistent with a 2010 issue). Cite-key year is 2010.
• No DOI. Reports of the National Food Research Institute are not routinely DOI-minted; doi: null with no_doi_assigned: true and a Google Scholar search access_url is the documented fallback pattern (see wiki/lint/2026-05-13-doi-fallback-followup.md). NARO maintains an institutional repository where the report PDF can be retrieved, but no stable DOI redirect or verified institutional-repository URL was identified during ingest. A future FIX-MISSING-DOIS pass may upgrade the access_url to the NARO repository URL if confirmed.
• soy-sauce and miso product slugs absent from taxonomy. Surfaced for Karen’s future taxonomy-review attention. Current routing uses [[products/condiments-general]] for the soy-sauce side and [[products/soy-products]] for the miso side. Per CLAUDE.md, this ingest skill does not create product pages; the gap is logged for the taxonomy review pass.
• Folder placement. PDF is filed under raw/manual-fetch/Kimi_Agent_Download Corruption Issue/condiments2_papers/03_Condiments/. The folder is the upstream sorter’s category and is preserved (immutable per Part 4). This is the fifth file in the 03_Condiments subfolder; companion sources from the same subfolder include rodriguez2009-fish-sauce-arsenic-speciation.md (02_Rodriguez) and bergerritchie2013-hot-sauce-lead.md (likely earlier in the numbered sequence).

Wiki pages this source may touch

• cadmium
• soy
• rice
• wheat
• condiments-general
• soy-products
• Future soy-sauce (does not yet exist; condiments-general is the current routing destination)
• Future miso (does not yet exist; soy-products is the current routing destination)
• Future soybean (does not yet exist; soy is the current routing destination)
• Future rice-koji (does not yet exist; rice is the current routing destination)

Page history

The five most recent substantive edits to this page. The full version history lives in git; when DOI minting comes online (see schema docs), each entry below will also link to a version-pinned DataCite DOI.