In vitro impact of storage media on extracted teeth: Systematic review and meta-analysis of adhesive, mechanical, and crack-related outcomes

Abstract

Aims:

This systematic review aimed to evaluate the influence of storage medium, moisture status, temperature, and duration on the in vitro outcomes of extracted human teeth, including bond strength, microleakage, fracture resistance, and imaging-detected artifacts.

Methods:

The study is reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guideline. In vitro and ex vivo comparative studies on extracted human teeth were eligible. Animal studies, simulations, and noncomparative reports were excluded. MEDLINE (PubMed), Embase, Scopus, Web of Science, Cochrane CENTRAL, ClinicalTrials.gov, and Google Scholar were searched from inception to October 15, 2025. Risk of bias was assessed using the Quality Assessment Tool for In vitro Studies in Dentistry (QUIN). Random-effects meta-analyses were performed where outcomes and contrasts were sufficiently comparable; otherwise, results were synthesized by contrast families. Continuous outcomes were expressed as mean differences (MDs) or standardized MDs and binary outcomes as odds ratios (ORs).

Results:

Fourteen studies met the inclusion criteria. Dehydration consistently increased the detection of crack-like artifacts compared with moist storage (pooled OR = 56.66). Microleakage was greater under less favorable storage conditions (pooled MD = 0.97).

Limitations:

The evidence base consisted of small laboratory studies with variable reporting.

Conclusions:

Preanalytical storage conditions substantially influence in vitro dental outcomes. Maintaining moisture prior to testing minimizes artifact formation, while formalin and prolonged aging protocols may bias bonding and leakage assessments.

Registration:

The protocol had been registered in the International Prospective Register of Systematic Reviews (PROSPERO; CRD420251179938).

INTRODUCTION

Extracted human teeth are widely used as experimental substrates in dental materials science and biomechanics. The interval between extraction and testing is managed with diverse storage and handling protocols that differ in moisture status, medium composition, temperature, and duration.[1] Evidence from longitudinal micro-computed tomography (micro-CT) demonstrates that dehydration promotes the development of dentinal defects, whereas sustained storage in water prevents new defects over extended follow-up.[2,3]

Adhesive and sealing outcomes are also sensitive to storage conditions of teeth. Previous studies have shown that solution type and duration can modify dentin permeability and bond behavior, with prolonged saline storage producing large changes in permeability and selective effects on bond strength, underscoring the risk of using saline as a default medium.[4,5]

In microleakage studies, teeth kept in formalin showed reduced dye penetration compared with freshly extracted controls, an effect attributed to chemical fixation of collagen; by contrast, storage in chloramine generated results comparable to fresh teeth and is therefore considered methodologically acceptable.[6] The influence of storage duration is likewise evident: refrigeration in chloramine for 48 days or longer increases microleakage, while short-term refrigeration and cryopreservation did not change leakage in the same experiments.[7]

Moisture control appears central to adhesive performance. Contemporary data show that preserving dentin moisture yields higher microshear bond strength than dry storage or thymol-based storage, despite similar microhardness across groups.[8,9] With respect to mechanical behavior, storage chemistry and time can affect fracture outcomes. Sodium hypochlorite storage is associated with lower fracture loads, and fracture resistance tends to decline with increasing storage time.[10] These observations indicate that storage and handling are not neutral preparatory steps; they are upstream determinants of outcomes used to evaluate materials and techniques.

Despite the routine use of extracted human teeth in dental materials research, storage and handling protocols remain highly heterogeneous across laboratories. This variability can shift measurable outcomes such as bond strength, microleakage, and fracture resistance and can also introduce artifacts, thereby confounding comparisons between materials or techniques and limiting reproducibility. A systematic synthesis is therefore needed to consolidate and critically appraise the experimental evidence, quantify effects, and distinguish testing scenarios.

A key gap addressed by this review is the absence of a unified, outcome-spanning synthesis that evaluates storage handling effects on teeth within a single framework. Across literature, there is limited quantitative pooling of comparable contrasts, particularly for artifact formation where dehydration-related effects appear large and consistent. There exists an uncertainty regarding storage parameters (moisture status, medium chemistry, temperature, and duration) behaving as plausible effect modifiers. The evidence in this field has inconsistent methodological safeguards and low quality of reporting in laboratory studies, creating the requirement for a structured analysis and cautious interpretation. This review bridges these gaps by combining QUIN-based risk-of-bias appraisal with outcome-specific synthesis strategies such as random-effects meta-analysis and contrast-family descriptive synthesis.

This systematic review and meta-analysis evaluate how storage and handling of extracted human teeth influence four predefined outcomes: fracture resistance, bond strength, microleakage, and the detection of artifacts or microcracks on imaging. Where designs and measures are comparable, effects are pooled quantitatively, moisture status, storage temperature, and storage duration are explored as potential effect modifiers, and risk of bias is assessed.

METHODS

This systematic review and meta-analysis was conducted and reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guidelines.[11] The protocol was prospectively registered with the International Prospective Register of Systematic Reviews (PROSPERO) (Registration ID: CRD420251179938).

Eligibility criteria

We prespecified inclusion and exclusion criteria using the Population Intervention Comparator Outcome Study design (PICOS) framework.

1. Population/specimens: In vitro or ex vivo studies on extracted human teeth (intact or restored), prepared and stored under defined conditions prior to testing

2. Interventions: Any storage or handling condition capable of influencing tooth structure or imaging outcomes, including liquid media such as distilled water, saline, thymol, formalin and water, or dry storage; temperature manipulation, including refrigeration or cryostorage; and related preimaging dehydration protocols

3. Comparators: Another storage or handling condition (e.g. dry vs. moist, different durations of storage), or a concurrent control group

4. Outcomes: At least one of the following, reported with extractable data:

   1. Fracture resistance

   2. Bond strength

   3. Microleakage

   4. Artifacts/microcracks on imaging or microscopy.

5. Study design: Laboratory (in vitro or ex vivo) comparative studies with parallel groups or matched designs.

Exclusions

1. Studies conducted on animal teeth

2. Case reports/series without a comparator

3. Finite-element/analytical simulations

4. Reviews

5. Editorials

6. Conference abstracts without usable data

7. Studies lacking sufficient outcome details for effect-size computation and nonhuman materials

8. No date restrictions were considered

9. Articles other than those in English.

Information sources

Bibliographic databases and trial and registry platforms were searched from inception to October 15, 2025. We also screened reference lists and citations of included studies and relevant reviews. Databases searched included Medical Literature Analysis and Retrieval System Online (MEDLINE) (via PubMed), Embase, Scopus, Web of Science Core Collection, and Cochrane Central Register of Controlled Trials (Cochrane CENTRAL). The registries and trial platforms searched were ClinicalTrials.gov. Additional sources included Google Scholar.

Deduplication and management of data

All records were exported to a citation manager (Zotero) and deduplicated using automatic and manual matching, i.e. title, authors, year, and Digital Object Identifier (DOI). Screening was conducted against the eligibility criteria at the title/abstract and full-text levels.

Search strategy

Database-specific strategies were developed with input from a domain expert and an information specialist, and we reported them in full to ensure reproducibility. No date limits were applied. The final searches were run on October 15, 2025. The core string for PubMed using keywords and Medical Subject Headings terms is shown in Table 1. A similar strategy was adapted for Embase, Scopus, and Web of Science using equivalent Emtree/keyword mappings. Given resource limitations for translation and the predominance of English-indexed laboratory materials studies in major databases, the review was restricted to English-language full texts; nevertheless, this restriction may have excluded relevant evidence.

Table 1.

Search strategy

Element	Search String
Population	(tooth OR teeth OR dentin OR dentine OR enamel OR “extracted teeth” OR “human tooth” OR “Tooth” [MeSH] OR “Dentin” [MeSH] OR “Tooth, Extracted” [MeSH])
Intervention and comparator	(storage OR storing OR “storage medium” OR preservation OR preserv* OR dehydrati* OR “air-dry” OR dry OR desiccation OR moist OR wet OR hydration OR water OR “distilled water” OR saline OR “saline solution” OR thymol OR formalin OR ethanol OR alcohol OR “chloramine-T” OR chloramine OR NaOCl OR “sodium hypochlorite” OR cryopreserv* OR freezing OR “frozen storage” OR refrigeration OR “cold storage” OR “Specimen Handling” [MeSH] OR “Cryopreservation” [MeSH] OR “Freeze Drying” [MeSH] OR “Preservation, Biological” [MeSH])
Outcomes	(“fracture resistance” OR “load to fracture” OR “fracture strength” OR “mechanical strength” OR “bond strength” OR microtensile OR μTBS OR microshear OR SBS OR microleakage OR “dye penetration” OR “marginal leakage” OR artifact* OR artefact* OR microcrack* OR “crack formation” OR “dentinal defects” OR “Dental Leakage”[MeSH] OR “Dental Bonding”[MeSH] OR “Dentin-Bonding Agents”[MeSH] OR “Fractures, Stress”[MeSH] OR “Microscopy”[MeSH] OR “Tomography, X-Ray Computed”[MeSH])
Study design	(“in vitro” OR “ex vivo” OR laboratory OR experimental OR bench OR micro-CT OR micro-CT OR μCT OR CBCT OR “cone beam” OR “In Vitro Techniques” [MeSH])

Selection process

Records were imported into a citation manager, deduplicated, and screened in two stages. Two reviewers (KK and SA) independently screened titles and abstracts against eligibility criteria and then assessed full texts of potentially relevant records. Disagreements were resolved by discussion between the two. Unresolved conflicts were adjudicated by a third reviewer. Inter-rater agreement statistics were not calculated. We emphasized independent duplicate screening, prespecified eligibility criteria, and calibration of screening decisions to ensure consistency. We also ensured full transparency of the selection process through PRISMA reporting to appraise the selection process and its yield.

Data collection process

Two reviewers independently extracted data using a piloted spreadsheet. We recorded study identifiers, specimen characteristics, storage/handling protocols, comparators, measurement methods, and outcomes. For continuous outcomes, we extracted group sizes, means, and standard deviations (or convertible data) for binary outcomes, events, and totals. Unit conversions and assumptions were documented, and the most complete report was used when multiple reports existed.

Data items and outcomes

Four outcome domains were prespecified. Bond strength (microtensile or shear) was extracted as means and standard deviations and converted to megapascals (MPa) where required. Effect sizes were calculated as mean or standardized mean differences (MDs). Microleakage scores were harmonized and analyzed as continuous/ordinal or dichotomous data. Fracture resistance was defined as maximum load to failure and analyzed as MDs in Newtons (N). Artifacts/microcracks were treated as binary outcomes and summarized as odds ratios (ORs) with a continuity correction for zero cells. For artifact/microcrack outcomes, we recorded the detection modality (micro-CT vs. microscopy/sectioning). Where reported, we extracted key imaging and classification details (resolution/voxel size and defect definitions). However, these parameters were not consistently reported across studies and therefore could not be standardized quantitatively. Accordingly, artifact outcomes were synthesized using a common binary endpoint (presence of ≥1 defect), and detection heterogeneity was considered in interpretation.

Risk of bias assessment

Risk of bias in in vitro studies was assessed using the Quality Assessment Tool for In vitro Studies in Dentistry (QUIN) across 12 domains (including clarity of aims, sample size calculation, sampling, comparison groups, methodological standardization, operator/assessor training, randomization/allocation, outcome measurement, blinding, statistics, and reporting). Three storage-specific checks were also recorded descriptively: adequacy of storage control (medium, concentration, temperature, and duration), hydration control during processing/testing, and validity of measurement methods (e.g. nondestructive imaging vs. sectioning). Domain-level ratings (low/moderate/high/unclear) were combined into an overall judgment using predefined rules (low = no high domains; high = ≥2 high domains or any high in a critical domain). Two reviewers independently applied QUIN using a calibrated form. The ratings were performed independently using prespecified decision rules with a brief calibration exercise prior to full appraisal. Calibration consisted of a pilot assessment of three studies to harmonize domain interpretations. We standardized what constitutes adequate protocol standardization, assessor calibration, and blinding in storage and handling experiments before formal rating of all included studies. Disagreements were resolved by consensus, with third-reviewer adjudication when required. Results were tabulated and displayed as traffic-light plots.

Data synthesis

Storage and handling was treated as a preanalytical exposure defined by five parameters: (i) hydration status (dry vs. moist), (ii) medium chemistry (water/saline; disinfectants such as chloramine-T, NaOCl, and thymol; fixatives such as formalin; and alcohol-based media), (iii) temperature (room temperature/refrigeration/cryostorage), (iv) duration (short-term vs. prolonged storage, analyzed using time-matched contrasts), and (v) additional processing steps (sterilization/autoclaving when present). For synthesis, studies were organized into prespecified contrast families reflecting these parameters (for example, dry vs. hydrated; disinfectant A vs. disinfectant B; and prolonged vs. short-term aqueous storage). Sensitivity analyses were planned around these a priori families and around key modifiers (duration, temperature, and measurement modality).

Studies were grouped by specimen and storage/handling characteristics, test methods, and outcome family. Only contrasts with extractable, harmonizable data were synthesized; time-matched contrasts were treated as separate comparisons. Continuous outcomes were converted to means and standard deviations using standard formulas; binary outcomes were extracted as events and totals. Random-effects meta-analysis (DerSimonian–Laird) was used by default, with heterogeneity quantified by Cochran’s Q statistic (Q), I² statistic (I²), and between-study variance (τ²). Prespecified subgroup and sensitivity analyses explored moisture status, storage duration, temperature, measurement modality, and alternative handling of zero cells and multiple time points. All meta-analyses and forest plots were generated in Review Manager (RevMan) version 5.4 (Cochrane Collaboration). For binary outcomes with zero-event cells (common in moist/wet groups), ORs were computed using a 0.5 continuity correction to enable estimation and maintain consistency across contrasts.

Given the expected heterogeneity in tested outcomes, a protocol was set that quantitative pooling would be conducted only when the outcome constructs were comparable and had a broadly similar storage temperature and time window), and testing protocols were similar to support interpretability. When these conditions were not met, findings were synthesized as contrast families, presenting study-level effect estimates without overall pooling.

RESULTS

Study selection

Database and register searches identified 1811 records (1603 – database and 208 – register). Before screening, 366 were removed (21 duplicates, 238 by automation tools, and 107 for other reasons). The remaining 1445 titles/abstracts were screened; 1257 were excluded. Of 188 full-text reports assessed (plus seven sought separately and five unretrieved), 14 studies (14 reports) met the inclusion criteria. The PRISMA 2020 flow diagram [Figure 1a] summarizes selection and study characteristics, as shown in Table 2.[12,13,14,15,16,17,18,19]

Figure 1.

(a) Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow diagram elucidating the identification of included studies. (b) Traffic-light plot summarizing the risk of bias of included studies using the QUIN tool of 360 assessment per-study, per-domain judgments (12 QUIN items)

Table 2.

Characteristics of included studies

Author, year	Type of study, tooth type/region	Storage media used, concentration, temperature, and duration	Type of processing	Bond strength	Microleakage	Fracture resistance	Artifacts
De-Deus et al., 2024[3]	In vitro experimental	Water versus dry	Nondestructive micro-CT; no sectioning; dry versus hydrated groups	-	-	-	Micro-CT; longitudinal; new defects only in dry; none in water
Shemesh et al., 2018[12]	In vitro experimental	Moist preimaging, then air-dry at room temperature (5 min to 24 h) or slow dry at 4°C (7 days)	2% NaOCl irrigation; transverse sectioning under water coolant; optical + micro-CT	-	-	-	Dehydration alone → new cracks (even without instrumentation)
Goodis et al., 1993[4]	In vitro experimental	70% ethanol, 10% formalin, DW, DW + thymol, PBS + thymol; 8 days, 15 days, 6 months	EDTA for smear removal; air-dry for 15 s; standardized 3 mm Ø bond	SBS (shear); area 3 mm diameter (Ø); units MPa	-	-	-
Camps et al., 1996[7]	In vitro experimental	Fresh; cryopreserved (~13 weeks); 0.5% chloramine-T at 4°C (12 days or ≥48 days)	Sectioning; light microscopy; SEM/EDX	-	Silver nitrate dye; 100 thermocycles; ↑ leakage after ≥48 days refrigeration; cryo ~ neutral	-	-
Panighi et al., 1997[13]	In vitro experimental	Cryopreservation with DMSO (brief prefreeze) versus fresh; stored for 1 week DW before test	Flat dentin; composite cylinders 3 mm Ø ×3 mm	SBS (shear); 0.5 mm/min; MPa; more cohesive dentin failure superficially (cryo)	-	-	-
Toledano et al., 2007[14]	In vitro experimental	Prebond: 0.5% chloramine-T at 4°C (<1 month); postbond aging: Dry versus DW versus mineral oil (24 h/3 months/12 months)	Sectioned into μTBS beams; SEM failure analysis	μTBS (tensile); ≈1 mm2; MPa	-	-	-
Haller et al., 1993[6]	In vitro experimental	Storage discussed/tested: 1% chloramine, 70% ethanol, 10% formalin, 0.1% thymol (≈3–4 weeks; 4°C for chloramine)	Sectioning; microscopy	-	Basic fuchsin scores (0–4); 1440 thermocycles; formalin ↑ leakage; chloramine acceptable	-	-
Mobarak et al., 2010[15]	In vitro experimental	Fresh ≤2 weeks (0.5% chloramine-T), 0.5% chloramine-T at 4°C ~2 years, dry storage ~2 years (rehydrated 2 weeks)	Sectioning/wet grinding; stereomicroscope failures	μTBS (tensile); beam≈0.9 mm2; MPa; mostly adhesive; no significant difference across storage	-	-	-
Hashim and Abbott, 2022[10]	In vitro experimental	10% formalin, 0.2% thymol/saline, 5.25% NaOCl, DW, frozen moist; 14/90/180 days	Sectioning for roots; kept hydrated; Instron testing	-	-	Compressive load to fracture: 0.5 mm/min; N; NaOCl worst; DW↓at 90–180 days	-
Arashiro et al., 2019[16]	In vitro experimental	Dry for 30 days versus autoclave + DW for 30 days versus 10% formaldehyde 14 days + DW 30 days	Transverse sections (low-speed diamond, coolant); microscope ×30	-	-	-	Cracks only after 30 days of drying; none in autoclave + DW or formalin + DW
Angermair et al., 2020[17]	In vitro experimental	Cryopreservation: 10% DMSO; −80°C versus−196°C, 2 weeks	PMMA embedding; cyclic loading (50k) then load-to-failure	-	-	Load at fracture (N): No difference between−80/−196°C	-
Yilmaz et al., 2023[8]	In vitro experimental	0.1% thymol, DW, dry; (duration not clearly stated)	-	μSBS (shear); MPa; (also dentin moisture and microhardness)	-	-	-
Da Silva et al., 2015[18]	In vitro experimental	Aging: DW, mineral oil, 1% NaOCl; 24 h and 15 days	Composite microcylinders	μSBS (shear); 0.5 mm/min; MPa	-	-	-
Jarahi et al., 2018[19]	In vitro experimental	Thymol versus chloramine (>6 months); autoclave subgroup	Autoclaving evaluated; mounting	SBS (shear); MPa; chloramine ↑ SBS; thymol preferable >6 months	-	-	-

Ø: Diameter, DW: Distilled water, DMSO: Dimethyl sulfoxide, EDX: Energy-dispersive X-ray analysis, EDTA: Ethylenediaminetetraacetic acid, MPa: Megapascals, PBS: Phosphate-buffered saline, PMMA: Polymethyl methacrylate, RT: Room temperature, SEM: Scanning electron microscopy, NaOCl: Sodium hypochlorite, SBS: Shear bond strength, μSBS: Micro-SBS, μTBS: Microtensile bond strength, CT: Computed tomography, ↑: Increase, ↓: Decrease

Full texts were mainly excluded for lacking experimental comparators, insufficient numerical data, overlapping datasets, or use of nonhuman materials or nonprespecified outcomes.

Risk of bias

We assessed risk of bias using the Quality Assessment Tool for In vitro Studies in Dentistry (QUIN). For each study, domain-level judgments were categorized as low, moderate, high, or unclear risk of bias. Per-study, per-domain judgments (12 QUIN items) are shown in the traffic-light matrix in Figure 1b.

The traffic-light plot showed mixed overall quality. Most studies had low risk for clearly stated aims, appropriate comparison groups, generally standardized methods, and acceptable statistics/reporting (mainly green in Domains 1, 4, 5, 11, and 12).

However, sample size calculation (Domain 2) was found to be high risk in nearly all the studies, and sampling technique, operator and assessor calibration, randomization/allocation, and blinding (Domains 3, 6, 7, 9, and 10) are often moderate, high, or unclear, indicating limited control of selection and detection bias. QUIN per-domain judgments with brief justifications of included studies are summarized in Supplementary Table 1a.

Supplementary Table 1a.

Risk of bias assessment via quality assessment tool for in vitro studies in dentistry summarizing per-domain judgments with justifications for the included studies

Study	D1 Aim	D2 sample size calculation	D3 sampling technique	D4 comparison group	D5 methodology/standardization	D6 operator calibration	D7 randomization/allocation	D8 outcome measurement	D9 assessor calibration	D10 blinding	D11 statistics	D12 results reporting
De-Deus, 2023	L: Aim explicit (dehydration artifacts)	H: No power/sample size calculation	M: Convenience teeth; selection not random	L: Clear dry versus hydrated groups	L: Storage protocol described and consistent	U: Operator training/repeatability not reported	U: Allocation method not stated	L: Nondestructive longitudinal micro-CT	U: Examiner calibration not reported	U: Blinding not stated	M: Limited inferential modeling/precision reporting	L: Outcomes and group data clearly presented
Shemesh, 2018	L: Aim explicit (dehydration-induced cracks)	H: No power/sample size calculation	M: Tooth sampling frame not random	L: Comparator conditions clearly defined	L: Protocol described (drying intervals)	U: Operator calibration not reported	U: Allocation approach not described	L: Micro-CT/optical assessment described	U: Reliability testing not reported	U: Blinding not stated	L: Statistical methods adequately reported	L: Results clearly and fully reported
Goodis, 1993	L: Aim explicit (storage vs. bond)	H: No power/sample size calculation	M: Convenience sampling/limited selection detail	L: Appropriate storage comparators	L: Bond protocol generally standardized	U: Operator calibration not reported	U: Allocation not described	L: SBS method described	U: Assessor calibration not stated	U: Blinding not stated	M: Limited reporting of variance/precision	L: Numeric outcomes reported by group
Camps, 1996	L: Aim explicit (storage/cold effects)	H: No power/sample size calculation	M: Sampling not clearly representative	L: Fresh versus stored contrasts defined	M: Some storage parameters incompletely controlled	U: Operator calibration not reported	U: Allocation not stated	L: Dye leakage protocol described	U: Assessor reliability not reported	H: No blinding; subjective leakage scoring risk	M: Limited precision estimates/sparse reporting	L: Main outcomes reported with group values
Panighi, 1997	L: Aim explicit (cryopreservation effects)	H: No power/sample size calculation	M: Sampling not random; limited frame details	L: Comparator groups appropriate	L: Protocol largely standardized	U: Operator calibration not reported	U: Allocation not described	L: SBS measurement described	U: Reliability testing not reported	U: Blinding not stated	M: Limited reporting of dispersion/precision	L: Outcomes reported adequately
Toledano, 2007	L: Aim explicit (chloramine vs. bonding)	H: No power/sample size calculation	L: Tooth selection criteria described	L: Comparator groups appropriate	L: Adhesive/bond protocol standardized	U: Operator calibration not reported	U: Allocation not described	L: μTBS protocol detailed	U: Calibration not reported	U: Blinding not stated	L: Appropriate tests and reporting	L: Results reported clearly
Haller, 1993	L: Aim explicit (storage and leakage)	H: No power/sample size calculation	M: Sampling not random/limited frame	L: Comparator storage groups defined	M: Some parameters insufficiently standardized	U: Operator calibration not reported	U: Allocation not described	M: Leakage scoring prone to subjectivity	U: Assessor reliability not reported	H: Blinding not reported; subjective scoring risk	M: Limited precision/dispersion reporting	L: Results generally reported per group
Mobarak, 2010	L: Aim explicit (storage and μTBS)	H: No power/sample size calculation	L: Tooth selection criteria described	L: Comparator groups appropriate	L: Protocol standardized	U: Operator calibration not reported	U: Allocation not described	L: μTBS method detailed	U: Calibration not reported	U: Blinding not stated	L: Statistics appropriately reported	L: Results clearly reported
Hashim and Abbott, 2022	L: Aim explicit (media/time vs. fracture)	H: No power/sample size calculation	L: Selection criteria/time windows described	L: Comparators appropriate (media/time)	L: Storage and loading protocol described	U: Operator calibration not reported	U: Allocation not described	L: Objective load-to-fracture testing	U: Calibration not reported	U: Blinding not stated	L: Statistics reported adequately	L: Results clearly reported
Arashiro, 2019	L: Aim explicit (drying vs. cracks)	H: No power/sample size calculation	M: Sampling frame not random	L: Dry versus wet comparators defined	M: Some protocol elements not fully standardized	U: Operator calibration not reported	U: Allocation not described	M: Crack detection susceptible to processing artifact	U: Reliability testing not reported	H: Blinding absent; subjective microscopy risk	M: Limited precision reporting	L: Outcomes reported per group
Angermair, 2020	L: Aim explicit (cryostorage temp)	H: No power/sample size calculation	M: Sampling not random; limited frame	L: Comparator cryo-temperature groups defined	L: Protocol described and consistent	U: Operator calibration not reported	U: Allocation not described	L: Objective fracture testing	U: Calibration not reported	U: Blinding not stated	M: Sparse precision/variance reporting	L: Results reported clearly
Yilmaz, 2022	L: Aim explicit (media/dry vs. μSBS)	H: No power/sample size calculation	U: Tooth source/selection insufficiently described	L: Comparator conditions defined	M: Some storage parameters incompletely detailed	U: Operator calibration not reported	U: Allocation not described	M: μSBS method described; some protocol details limited	U: Reliability not reported	U: Blinding not stated	M: Limited variance/precision reporting	L: Outcomes reported per group
Silva, 2015	L: Aim explicit (media/aging vs. μSBS)	H: No power/sample size calculation	M: Sampling not random/limited frame	L: Comparator media/time groups defined	L: Protocol standardized	U: Operator calibration not reported	U: Allocation not described	L: μSBS testing described	U: Assessor calibration not reported	H: Blinding not described (operator awareness possible)	L: Statistical reporting adequate	L: Results clearly reported
Borouziniat/Jarahi, 2018	M: Aim stated but not sharply framed	H: No power/sample size calculation	U: Sampling frame/selection unclear	L: Comparator groups present	M: Method/protocol partially specified	U: Operator calibration not reported	H: Nonrandom or clearly biased allocation	M: Measurement method incompletely detailed	U: Assessor reliability not reported	H: Blinding absent/not feasible; subjective elements	M: Statistics incompletely reported	M: Incomplete outcome reporting/precision

SBS: Shear bond strength, μSBS: Micro-SBS, μTBS: Microtensile bond strength, L: Low risk, M: Moderate risk, H: High risk, U: Unclear

Meta-analysis

Bond strength

Bond strength outcomes were generated under markedly different storage mechanisms (pure dehydration vs. aqueous storage vs. chemical disinfection vs. prolonged aging), test methods (SBS/µSBS/µTBS), and protocols. Therefore, an overall pooled meta-analysis was not considered clinically or methodologically interpretable due to high heterogeneity. Hence, bond strength findings were synthesized as contrast families, reporting study-level MDs in megapascals (MPa) with 95% confidence intervals (CIs) for each prespecified comparison.

Microleakage

Two studies[6,7] reported dye penetration microleakage on comparable ordinal scales, comparing stored versus fresh teeth (chloramine at 4°C ≥48 days vs. fresh) and chloramine versus formalin (~3–4 weeks). For Haller et al.,[6] dye penetration depth was originally recorded as an ordinal 0–4 score with explicit mm cutoffs (0 = 0 mm; 1= <0.25 mm; 2 = 0.25–0.5 mm; 3 = 0.5–<1 mm; and 4= >1 mm). The plotted mm values were derived from the study’s 0–4 dye penetration score using the prespecified score-to-mm midpoint conversion. We reconstructed approximate mm values by assigning each score category the midpoint of its mm interval (0, 0.125, 0.375, 0.75 mm).

The random-effects meta-analysis (MD in arbitrary scale units) showed +0.97 (95% CI: 0.59–1.35; I² = 43.1%), indicating consistently greater leakage with the less favorable storage condition.

Haller et al.[6] found minimal microleakage (~0.20 mm) in freshly extracted teeth with Scotchbond Multi-Purpose and Gluma 2000, whereas Camps et al.[7] reported higher values (0.53–1.00 mm) across five dentin bonding agents under stored conditions. Freshly extracted teeth showed lower microleakage than stored specimens, underscoring the combined influence of adhesive formulation and storage on marginal integrity, as depicted in Figure 2a.

Figure 2.

(a) Forest plot comparing mean microleakage values between studies by Camps et al.[7] and Haller et al.[6] (b) Forest plot comparing mean fracture resistance values between studies by Hashim and Abbott[10] and Angermair et al.[17] (c) Forest plot comparing odds ratio of presence of artifacts between studies by De-Deus et al.,[3] Shemesh et al.,[12] and Arashiro et al.[16]

Fracture resistance

Contributing studies[10,17] evaluated monotonic load-to-fracture in extracted human teeth, mainly comparing NaOCl with distilled water (DW) at 14, 90, and 180 days and −80°C with −196°C cryostorage. Loading protocols were broadly similar but not identical. QUIN ratings were mixed, with frequent issues in blinding/operator calibration and incomplete reporting of storage control (medium, temperature, and refresh schedules).

Fracture resistance evidence was sparse and dominated by a single study contributing multiple time-point contrasts. Accordingly, we present time-matched study-level MDs (14/90/180 days) and interpret these descriptively. For NaOCl versus DW (k = 3 time-matched contrasts from Hashim and Abbott;[10] n = 10 per group at each interval), a random-effects meta-analysis using mean difference (MD) in Newtons (N) showed an effect estimate of −84.5 (95% CI: −241.3 to 72.2; I² = 15%). This suggests a small and imprecise reduction in fracture resistance with NaOCl, with no clear overall difference between the groups. Leave-one-out analyses did not change the direction of effect and remained nonsignificant (pooled MD range: −142.0 to −16.7 N; all 95% CIs crossing 0). Restricting to the 180-day interval yielded MD −84.1 N (95% CI: −315.7 to 147.5), consistent with the main analysis.

The cryostorage comparison (−80°C vs. −196°C; Angermair;[17] k = 1, n = 10 per arm) suggested lower mean fracture loads at −80°C (MD: −310.0 N; 95% CI: −743.6 to 123.6), but with wide, nonsignificant CIs. Overall, trends favored DW over NaOCl and −196°C, over −80°C, without statistically significant differences. A forest plot of fracture resistance is shown in Figure 2b.

Artifacts

Three studies[3,12,16] compared dry/air-dry with moist/wet storage, assessing artifacts/microcracks at the tooth/section level by imaging or microscopy. Protocols were broadly comparable; the main risk of bias concerns assessed by QUIN tool were related to inadequate control of hydration during specimen handling and the possibility of artifact formation during destructive sectioning procedures. It is important to note that detection modality differed across studies (3D volumetric micro-CT vs. 2D microscopy/sectioning). To standardize outcomes, we used the most comparable endpoint available across studies by the presence of ≥1 defect/crack (binary) and extracted events and totals accordingly.

Across De-Deus et al.[3] (dry vs. water: 16/20 vs. 0/20), Shemesh et al.[12] (air-dry vs. moist: 6/10 vs. 0/10), and Arashiro et al.[16] (dry vs. wet pooled: 3/10 vs. 0/40), all events occurred in the dry/air-dry arms, with zero events in the moist/wet groups. Using a uniform 0.5 continuity correction, the random-effects meta-analysis gave a pooled OR of 56.66 (95% CI: 9.77–328.75; I² = 0%, τ² = 0), indicating a large, consistent effect favoring moist/wet storage. Leave-one-out analyses (k = 3) yielded similarly large ORs (~32.8–107.0), all with 95% CIs >1 and I² = 0%. These results show a robust association between dehydration and detectable artifacts, consistent with the expectation that loss of moisture accentuates or propagates microdefects and makes them more visible on imaging [Figure 2c].

Outcome-specific effect estimates for bond strength, microleakage, fracture resistance, and artifact formation are summarized in Supplementary Tables 1b-e.

Supplementary Table 1b.

Bond strength findings reported as mean differences across each group in a study (mean differences, MPa)

Contrast family	Study	Effect (MD, MPa), 95% CI	Interpretation
Hydration versus dehydration	Yilmaz et al.[2]	Water versus dry: +21.56 (20.19 to 22.93)	Distilled water produced substantially higher bond strength than dry storage
Alcohol-based storage versus water	Goodis et al.[4]	Ethanol/alcohol versus water: −2.60 (−4.78 to−0.42)	Alcohol-based storage reduced bond strength compared with water (small-to-moderate magnitude)
Oxidizing disinfectant versus water	Da Silva et al.[18]	NaOCl versus water: −2.10 (−4.72 to 0.52)	Trend towards lower bond strength with NaOCl; CI crosses zero (imprecise)
Chloramine versus thymol	Jarahi et al.[19]	Chloramine-T versus thymol: +0.77 (−0.16 to 1.70)	Small difference; compatible with no effect
Long-term dry versus long-term chloramine	Mobarak et al.[15]	Dry versus chloramine-T: +2.70 (−4.31 to 9.71)	Uncertain direction; wide CI indicates low precision for this contrast
Aging duration in water	Toledano et al.[14]	Water (12 months) versus 24 h: −15.61 (−19.42 to−11.80)	Prolonged aqueous aging markedly reduced bond strength compared with short-term storage

CI: Confidence interval, NaOCl: Sodium hypochlorite, MD: Mean difference

Supplementary Table 1e.

Artifact/microcrack detection reported as odds ratios across each group (odds ratio)

Contrast family	Study	Effect (OR, 95% CI)	Interpretation
Dry versus moist storage	De-Deus et al.[3]	Dry versus water: OR → ∞ (continuity-corrected OR >50)	Artifacts occurred exclusively after dehydration; none observed in hydrated teeth
Air-dry versus moist	Shemesh et al.[12]	Air-dry versus moist: OR → ∞	Dehydration alone induced new cracks, even without instrumentation
Long-term dry versus wet	Arashiro et al.[16]	Dry versus wet: OR → ∞	Prolonged dry storage strongly associated with crack formation

All ORs calculated with a 0.5 continuity correction due to zero-event cells in moist storage groups. CI: Confidence interval, OR: Odds ratio

Supplementary Table 1c.

Microleakage findings reported as mean differences across each group in a study (mean differences, mm)

Contrast family	Study	Effect (MD, 95% CI)	Interpretation
Stored versus fresh teeth	Camps et al.[7]	Stored in chloramine for ≥48 days versus fresh: +0.97 (0.59 to 1.35)	Prolonged refrigerated storage increased microleakage compared with fresh teeth
Disinfectant type	Haller et al.[6]	Formalin versus chloramine: Positive MD (direction favors chloramine)	Formalin storage artifactually reduced dye penetration, potentially overstating sealing performance
Cryopreservation versus fresh	Camps et al.[7]	Cryopreserved versus fresh: ≈0	Cryopreservation did not meaningfully alter microleakage compared with freshly extracted teeth

CI: Confidence interval, MD: Mean difference

Supplementary Table 1d.

Fracture resistance findings reported as mean differences across each group in a study (mean differences, n)

Contrast family	Study	Effect (MD, n, 95% CI)	Interpretation
Oxidizing disinfectant versus water	Hashim and Abbott[10]	NaOCl versus DW: −84.5 (−241.3 to 72.2)	Small, imprecise reduction in fracture resistance with NaOCl; no clear difference overall
Storage duration (DW)	Hashim and Abbott[10]	180 days versus 14 days: Negative MD (direction favors 180 days)	Longer storage tended to reduce fracture resistance, but estimates were imprecise
Cryostorage temperature	Angermair et al.[17]	−80°C versus−196°C: −310.0 (−743.6 to 123.6)	No statistically significant difference between cryostorage temperatures

MD: Mean difference, DW: Distilled water, CI: Confidence interval, NaOCl: Sodium hypochlorite

Assessment of publication bias was planned only when ≥10 studies were available per meta-analysis, consistent with guidelines that such methods are unreliable with sparse evidence. As the quantitative syntheses in this review included few studies per outcome, formal assessments of publication bias were not performed.

DISCUSSION

Consistent previous research shows that how extracted teeth are stored and handled between extraction and testing, particularly the choice of medium, temperature, and duration, can meaningfully alter mechanical, interfacial, and imaging outcomes. Dehydration tends to accentuate or reveal crack-like defects on imaging, while moist conditions help preserve tissue integrity. Storage chemistry and time can also shift bond strength and microleakage in directions that are not always clinically representative. Since these preanalytical factors act upstream of every in vitro measurement, they can confound comparisons across studies and complicate the interpretation of pooled results.

This review examined how preanalytical storage of extracted human teeth influences experimental outcomes that are commonly used to assess clinical performance. Across outcomes, the clearest and most consistent signal emerged for artifact formation. Dehydration markedly increased the odds of visible microcracks or related defects on imaging, whereas maintaining moisture essentially abolished new defect detection. The pooled odds ratio indicating greater artifacts under dry versus moist conditions was large with negligible heterogeneity, and this pattern was concordant with individual reports in which no new defects were observed when teeth were stored in water before scanning. These observations are biologically plausible because desiccation reduces dentin toughness and accentuates preexisting craze lines, while moist storage preserves the collagen–mineral interface and dampens crack propagation.[3,12,16]

Complementing these storage-related findings, Damor et al.[20] showed that targeted endodontic microsurgery using a trephine drill produced significantly more complete root-end microcracks than freehand apicoectomy with a carbide bur, emphasizing that surgical instrumentation can itself generate defects that may compound storage-related artifacts. Micro-CT work by Singh et al.[21] further demonstrated that root canal preparation, obturation, and especially post space preparation with ParaPost drills significantly increased the percentage of dentinal microcracks, with some cracks attributable to preexisting noniatrogenic factors such as functional loading, extraction forces, or dehydration.

For microleakage, the direction of effect depended on medium and duration. Dye penetration work showed that formalin can artifactually reduce leakage, likely via aldehyde-mediated collagen cross-linking, thus overstating adhesive performance relative to fresh dentin. On the other hand, 1% chloramine tended to approximate freshly extracted controls. In contrast, prolonged refrigeration in chloramine at 4°C beyond about 7 weeks was associated with higher leakage, while short-term refrigeration or cryopreservation appeared neutral. Together, these data suggest that “benign” media can become problematic with extended storage times and that formalin should be avoided when leakage or bonding is the endpoint.[6,7] Attam et al.[9] reported that 2-week storage in 10% formalin significantly reduced dye penetration compared with saline controls, whereas autoclaving produced microleakage values comparable to the control group, illustrating how sterilization and disinfection protocols can themselves bias leakage measurements.

Evidence for bond strength remained heterogeneous. Scoping syntheses and individual experiments reported mixed effects of common media. Thymol and ethanol were associated with lower values in some settings, 1% chloramine often tracked close to fresh teeth, and extended water storage after bonding could depress microtensile strength compared with dry or oil storage in long-term aging designs. Variability in adhesive systems, substrate depth, and postbond aging protocols likely contributes to dispersion in effects and limits straightforward pooling.[4,13,14,22] In a crown-fragment reattachment model, Shirani et al.[23] found that specimens stored in milk or saliva for 24 h exhibited higher fracture forces at debonding than those stored in water, normal saline, or in a dry environment, underlining the importance of maintaining coronal fragments in hydrated, biologically compatible media before bonding.

For fracture resistance, pooled contrasts suggested no clear, statistically significant difference between sodium-hypochlorite-exposed teeth and distilled water controls at matched time points, and wide CIs around single-study cryostorage comparisons precluded firm conclusions. Point estimates favored higher resistance in water and at lower cryogenic temperatures, but imprecision indicates that any true effect is likely small to modest at most within the tested ranges.[10] Consistent with this, Prabhakar et al.[5] showed that fragment reattachment after storage in milk, coconut water, or egg white yielded greater fracture resistance than fragments stored in dry air, supporting the use of readily available natural media to preserve mechanical performance before laboratory testing.

Sensitivity analyses did not materially change the direction or interpretation of pooled results. In leave-one-out analyses, removal of any single study/contrast yielded pooled estimates that were consistent with the main models for both fracture resistance and artifact outcomes. For fracture resistance, repeating the analysis using a single time point to minimize within-study nonindependence produced conclusions consistent with the primary synthesis. For sparse binary outcomes with zero-event cells, results remained stable under the prespecified continuity correction approach.

A further methodological consideration is that destructive processing such as sectioning, polishing, and dye penetration protocols can itself introduce microdefects or alter substrate characteristics, thereby influencing apparent microleakage and artifact prevalence. Where available, nondestructive longitudinal imaging such as micro-CT should be used. It provides stronger internal validity for distinguishing newly formed defects from preexisting features and from preparation-induced artifacts.

Limitations

Most contributing studies in this review were small bench experiments with variable reporting of allocation, blinding of assessors, and standardization of storage timing and temperature. Outcomes such as dye-based leakage and sectioned-specimen crack counts are prone to measurement bias and may not scale linearly to clinical failure modes. The synthesis relied on study-level summaries from a diverse set of protocols. Some meta-analyses required imputation or continuity corrections for zero-event cells in artifact outcomes, and unit-of-analysis issues may persist where multiple specimens per tooth were reported without clear clustering adjustments. The findings of this study should be interpreted in light of potential selection effects arising from the English-language restriction and the absence of a systematic gray literature search.

CONCLUSIONS

Preanalytical storage and handling of extracted teeth meaningfully influence laboratory outcomes and can introduce bias if left uncontrolled. Across the included evidence, dehydration consistently and markedly increases the detection of crack-like artifacts on imaging, whereas maintaining moisture essentially abolishes new defect formation, reinforcing hydration as a nonnegotiable prerequisite before imaging or sectioning.

These findings support routine use of moist storage (water or a hydration-preserving disinfectant) prior to imaging, avoidance of formalin for studies of microleakage or bonding, and careful limitation of storage duration when chloramine is used. Future studies should aim at standardized reporting of medium, temperature, pH, and precise storage intervals, along with preregistered protocols and clustering-aware analyses to strengthen future syntheses and improve reproducibility.

Conflicts of interest

There are no conflicts of interest.

Funding Statement

Nil.