Ethical considerations

This study was approved by the Institutional Ethics Committee of Douala Gynaeco-Obstetric and Pediatric Hospital (DGOPEH), Cameroon (approval number: N°2024/1617/HGOPED/DG/CEI). Prof. Emile Telesphore MboudoU, Chair of the DGOPEH Institutional Ethics Committee, was not part of the research team at the time of protocol submission and ethics review; he therefore had no conflict of interest with respect to this protocol during its evaluation by the committee. He was subsequently associated with the preparation of the present manuscript in his institutional capacity (Project administration, Resources). The study utilized surplus surgical waste tissue from a mastectomy specimen accessed on 30/03/2023. The specimen was completely anonymized prior to analysis and identified only by laboratory code assigned during routine pathology processing. Authors had no access to information that could identify the participant during or after data collection. Data collection and analysis were performed between March 2023 and March 2026.

Study design and specimen characteristics

We conducted an exploratory, proof-of-concept controlled experimental study at the Anatomic Pathology Department of DGOPEH, Cameroon, between March and July 2023. The biological material consisted of a fresh mastectomy specimen with axillary lymph node dissection accessed on 30/03/2023 from a 34-year-old female patient diagnosed with invasive ductal carcinoma of the left breast who had not received neoadjuvant chemotherapy. A single-specimen experimental design was chosen to eliminate inter-tumor biological variability as a confounding factor, consistent with established methodology in pre-analytical fixation research [8,11,17]. Pre-operative core needle biopsy had established moderately differentiated invasive ductal carcinoma, clinical stage pT4, positive for ER and PR, negative for HER2 by IHC, and highly proliferative with a Ki67 index of 40%.

Experimental design

Statistical analysis

Statistical analyses were conducted using IBM SPSS Statistics version 25.0 and Python (SciPy 1.11). Continuous variables were presented as mean ± standard error of the mean (SEM). Spearman’s rank correlation coefficient (ρ) was used to evaluate the relationship between fixation duration or cold ischemia time and biomarker expression levels. Paired t-tests were utilized to compare mean biomarker expression levels between the two fixative types, with 95% confidence intervals (CI) calculated for mean differences. To formally test the effect of fixation duration, we also performed one-way ANOVA comparing four a priori fixation-duration windows (0.5–12 h, 24–48 h, 72–96 h, 120–144 h) separately for each fixative and each biomarker, with Welch pairwise t-tests between the reference window and the other windows. Two-way ANOVA was performed to assess interaction effects. A two-tailed p-value < 0.05 was considered statistically significant. Interobserver variability was evaluated by calculating the intraclass correlation coefficient (ICC). We note that, given the single-specimen design, reported p-values describe within-specimen technical variation under controlled pre-analytical conditions and do not support population-level inference.

Sample characteristics and quality control

All 50 samples contained tumor tissue as confirmed by H&E staining. The ICC between the two independent observers was 0.94 for ER, 0.92 for PR, and 0.96 for Ki67 evaluation, indicating excellent inter observer agreement. Under optimal conditions (immediate fixation in 4% NBF for 0.5–12 hours), the tumor exhibited strong biomarker expression: ER 96–100% positive cells (Allred score 7–8), PR 95–100% positive cells (Allred score 7–8), and Ki67 38–40% (high proliferative).

Effect of fixative preparation on biomarker expression

When comparing samples fixed in 4% NBF versus 4% NB-F across all fixation durations (0.5–144 hours), statistically significant differences were observed in both the percentage of positive cells and staining intensity for ER and PR (Table 1).

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Comparison of biomarker expression between fixative types.

https://doi.org/10.1371/journal.pone.0343185.t001

Despite statistical significance, Allred scores remained unchanged between the two fixative types for fixation durations up to 48 hours. Both fixatives showed progressive decline in expression after 48 hours (Fig 1).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Effect of fixation duration on biomarker expression in invasive breast cancer.

(A) Estrogen receptor (ER) percentage of positive cells. (B) Progesterone receptor (PR) percentage of positive cells. (C) Ki67 proliferation index. In each panel, samples fixed in 4% neutral buffered formaldehyde (NBF, blue) are overlaid with samples fixed in 4% non-buffered formaldehyde (NB-F, magenta). Data points represent mean values from duplicate observations by two independent evaluators. Note logarithmic x-axis scale. Biomarker expression remained stable up to 48 hours, followed by progressive decline with extended fixation (72–144 hours). Both fixative types showed similar preservation patterns, with slightly higher values for NBF (p < 0.05).

https://doi.org/10.1371/journal.pone.0343185.g001

Effect of fixation duration on biomarker expression

For samples with immediate fixation (no cold ischemia), biomarker expression remained optimal and stable from 0.5 to 48 hours for both fixatives, with Allred scores remaining at 7–8 for hormone receptors and Ki67 at 38–40%. After 48 hours, gradual decline was observed, with more pronounced decrease after 72 hours (Fig 2). Spearman rank correlations between fixation duration (0.5–144 h) and percentage of positive cells were statistically significant for all three biomarkers in both fixatives: ER — NBF ρ = −0.736 (p = 0.0003), NB-F ρ = −0.658 (p = 0.0022); PR — NBF ρ = −0.738 (p = 0.0003), NB-F ρ = −0.830 (p < 0.0001); Ki67 — NBF ρ = −0.596 (p = 0.007), NB-F ρ = −0.646 (p = 0.003). One-way ANOVA across fixation-duration windows (0.5–12 h, 24–48 h, 72–96 h, 120–144 h) was significant for all biomarkers in both fixatives: ER — NBF F = 22.96 (p < 0.0001), NB-F F = 25.63 (p < 0.0001); PR — NBF F = 28.99 (p < 0.0001), NB-F F = 45.48 (p < 0.0001); Ki67 — NBF F = 5.93 (p = 0.007), NB-F F = 6.13 (p = 0.006). Pairwise Welch t-tests indicated that the 24–48 h window was not significantly different from the 0.5–12 h reference window (all p > 0.15), whereas the 120–144 h window differed significantly for ER and PR percentages in both fixatives. At extended fixation durations (Table 2), progressive deterioration occurred with both fixatives.

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Biomarker expression at extended fixation times (ER %/ PR %/ Ki67%, NBF vs NB-F).

https://doi.org/10.1371/journal.pone.0343185.t002

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Effect of fixation duration on staining intensity.

(A) Estrogen receptor staining intensity. (B) Progesterone receptor staining intensity. In each panel, NBF (blue) and NB-F (magenta) are overlaid. Intensity scores: 0 = none, 1 = weak, 2 = moderate, 3 = strong. Both fixatives maintained strong to moderate intensity (scores 2.5–3.0) for hormone receptors up to 48 hours, with gradual decline to moderate intensity (scores 1.5–2.0) at extended fixation times (120–144 hours).

https://doi.org/10.1371/journal.pone.0343185.g002

Effect of cold ischemia time on biomarker expression

Cold ischemia time demonstrated strong negative correlation with biomarker expression, regardless of fixative type (Fig 3). For ER, expression remained near baseline (95–97% positive cells, intensity 2.5–3.0) at 0.5–1 hour, showed noticeable decline at 2 hours (79–83% positive, intensity 2.0–2.5), significant decrease at 4 hours (61–63% positive, intensity 1.5–2.0), marked reduction at 6 hours (41–52% positive, intensity 1.0–1.5), and severe decline at 8 hours (33–35% positive, intensity 1.0). Spearman correlations were ρ = −0.95 (p < 0.001) for NBF and ρ = −0.97 (p < 0.001) for NB-F.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Effect of cold ischemia time on biomarker expression.

(A) Estrogen receptor percentage of positive cells. (B) Progesterone receptor percentage of positive cells. (C) Ki67 proliferation index. In each panel, samples fixed in NBF (blue) and NB-F (magenta) after the indicated cold ischemia time are overlaid; the dashed vertical line marks the 2-hour threshold beyond which Allred scores began decreasing significantly. Strong negative correlations were observed for all biomarkers (Spearman ρ ≤ −0.94, p < 0.001). Biomarker expression remained adequate (>90% for hormone receptors, 38–40% for Ki67) up to 1 hour. Noticeable decline occurred at 2 hours, with dramatic deterioration beyond 4 hours, regardless of fixative type. All samples underwent standardized 10-hour fixation after the specified cold ischemia time. NBF, Neutral buffered formaldehyde; NB-F, Non-buffered formaldehyde.

https://doi.org/10.1371/journal.pone.0343185.g003

PR expression demonstrated similar sensitivity with even steeper decline: baseline at 0.5–1 hour (95–97% positive, intensity 2.5–3.0), decline to 80–81% at 2 hours, 59–66% at 4 hours, 47% at 6 hours, and 24–31% at 8 hours (some becoming negative). Spearman correlations were ρ = −0.96 (p < 0.001) for NBF and ρ = −0.98 (p < 0.001) for NB-F.

Ki67 expression also declined significantly: 38–40% at 0.5–1 hour, 38–39% at 2 hours, 33–35% at 4 hours, 27–31% at 6 hours, and 9–21% at 8 hours. Spearman correlations were ρ = −0.94 (p < 0.001) for NBF and ρ = −0.96 (p < 0.001) for NB-F.

Analysis revealed critical time thresholds: ≤ 1 hour showed no significant Allred score changes; 2 hours showed borderline changes but most samples remained positive with slightly reduced scores; > 2 hours showed progressive and clinically significant decline with potential for false-negative results.

Two-way ANOVA revealed cold ischemia time as the dominant factor (F = 87.3, p < 0.001), with no significant effect of fixative type (F = 2.1, p = 0.156) or interaction (F = 0.8, p = 0.523).

Principal findings and fixative preparation

We found three main results. First, formaldehyde preparation type (neutral buffered versus non-buffered) showed statistically significant but small differences in this index tumor. Although 4% NBF demonstrated slightly superior preservation of ER (mean difference 2.58%, p = 0.011) and PR (mean difference 2.26%, p = 0.027) compared to 4% NB-F, these differences did not translate into Allred score changes or clinical classification changes in our strongly ER- and PR-positive index tumor. We emphasize, however, that these differences are of a magnitude that could cross the diagnostic threshold in tumors with weak baseline expression (e.g., ER-low-positive tumors with 1–10% positive cells, as defined in the ASCO/CAP 2020 update [4]), and our data therefore do not support a general claim of equivalence between the two fixative preparations. Second, both fixative types adequately preserved biomarkers for 0.5 to 48 hours, after which progressive deterioration occurred. Third, cold ischemia time emerged as the dominant pre-analytical factor, with a critical threshold of approximately 2 hours beyond which clinically relevant biomarker degradation occurred regardless of fixative type. Although each of these effects individually replicates prior work, the contribution of the present study lies in their joint evaluation under controlled conditions on material from a single tumor.

The minimal difference between neutral buffered and non-buffered formaldehyde is somewhat counterintuitive given pH-controlled fixation’s theoretical advantages. Formaldehyde fixation proceeds through complex reactions involving formaldehyde addition to amino groups, followed by dehydration and methylene bridge formation [10]. While neutral buffered formaldehyde at pH 6.8–7.2 is optimal for initial addition, non-buffered formaldehyde (acidic through formic acid formation: H-CHO + H₂O → H-COOH + H₂) may facilitate the dehydration step more efficiently due to lower pH (~4.0–5.5) [10,23].

Our findings align with Apple et al., who found different fixative types yielded comparable hormone receptor results to standard neutral buffered formalin [17], and Oyama et al., who reported fixative type did not significantly influence ER assessment with automated IHC protocols [18]. Our study extends these findings by specifically comparing the most commonly available formaldehyde preparations in resource-limited settings and demonstrating that statistical differences do not translate into clinically meaningful changes in a strongly ER/PR-positive tumor.

This finding is particularly relevant for pathology laboratories in LMICs where neutral buffered formalin may be expensive, unavailable, or subject to supply chain disruptions. Our results suggest laboratories using properly diluted non-buffered formaldehyde (prepared from commercial 37–40% formalin) can achieve adequate biomarker preservation for routine IHC analysis, provided other pre-analytical variables, particularly cold ischemia time, are carefully controlled, and that this interpretation is confirmed in multi-specimen studies across diverse tumor types.

Effect of fixation duration

Biomarker expression in our specimen remained stable throughout the 0.5–48 h fixation window and declined progressively at extended durations, with statistically significant one-way ANOVA effects across fixation windows (all F ≥ 5.93, p ≤ 0.007) and strong negative Spearman correlations between fixation duration and percentage of positive cells (ρ between −0.60 and −0.83, p ≤ 0.007). The stability observed up to 48 h is consistent with the ASCO/CAP-recommended 6–72 h window [4,5] and with prior systematic reviews showing that hormone receptor and HER2 expression are well preserved within this interval [19]. The progressive decline observed beyond 72 h, most marked at 120–144 h, is in line with earlier experimental reports of gradual antigen loss with prolonged formalin exposure [11,12] and likely reflects over-cross-linking of protein epitopes and subtle tissue-matrix changes that compromise immunoreactivity. Notably, PR percentages declined earlier and more steeply than ER percentages, whereas Ki67 was relatively preserved across the full 0.5–144 h range — a pattern consistent with the differential protein stability of these three antigens under prolonged fixation. The clinical relevance of these observations is that fixation durations of 72–144 h, although outside the ASCO/CAP-recommended window, are empirically observed in sub-Saharan African pathology laboratories — particularly for mastectomy specimens, which routinely reach 43–147 h of fixation at our institution, but also for biopsies, which may reach up to 83 h [22], when specimens are held in fixative over weekends or public holidays. Our data suggest that in strongly-positive tumors such occasional prolongation may not change categorical classification, but also that this reassurance may not extend to tumors with weaker baseline expression, where the observed absolute decline could cross the diagnostic threshold.

Critical impact of cold ischemia time

The most clinically significant finding is cold ischemia time’s profound negative impact on biomarker preservation. We observed strong negative correlations (Spearman ρ ≤−0.94, p < 0.001) between cold ischemia time and all tested biomarkers, with dramatic effects beyond 2 hours. At 8 hours, PR expression declined to borderline negative levels (24–31%), and Ki67 dropped from high proliferative (38–40%) to low proliferative (9–21%) ranges. These changes have direct clinical implications, potentially leading to inappropriate treatment decisions such as withholding endocrine therapy or underestimating tumor aggressiveness, a concern recently echoed in the context of post-neoadjuvant breast cancer specimens [21].

Our findings partially contradict Apple et al. [17], but this discrepancy may be explained by differences in initial biomarker expression levels. Several studies demonstrated that cold ischemia time’s impact is inversely related to baseline expression intensity [13,14]. Tumors with strong initial positivity (>80% cells with strong intensity), like our specimen, may tolerate brief fixation delays with minimal classification changes. However, tumors with moderate or weak baseline expression (10–50% positive cells or weak-moderate intensity) are more vulnerable to cold ischemia-induced degradation and may shift from positive to negative status with relatively short delays [11,12].

The mechanistic basis relates to continued tissue protease and degradative enzyme activity following tissue excision. While tissue no longer receives blood supply, cellular metabolism and enzymatic activity continue for hours, particularly for abundant enzymes such as proteases, phosphatases, and nucleases [9]. Formaldehyde fixation rapidly inactivates these enzymes by protein cross-linking, “freezing” the tissue’s molecular state. Delayed fixation allows progressive enzymatic antigen degradation, with particularly rapid effects on labile proteins [11,24].

Our observation that PR showed greater cold ischemia sensitivity than ER is consistent with previous reports [12] and has important clinical implications. PR expression is considered a functional indicator of ER pathway integrity, and ER + /PR- tumors may represent a distinct biological subtype with different clinical behavior [25]. Preferential PR loss with delayed fixation could lead to misclassification of true ER + /PR+ tumors as ER + /PR-, potentially affecting treatment decisions and prognostic assessments.

Practical implications and study limitations

If confirmed in multi-specimen studies, our findings would have potentially important practical implications for pathology practice in resource-limited settings. First, they would suggest that laboratories lacking access to commercial neutral buffered formalin may be able to use properly prepared non-buffered formaldehyde without substantially compromising hormone receptor and Ki67 assessment quality in strongly-positive tumors, pending validation in tumors with borderline expression. Such a finding could help reduce costs and improve diagnostic service sustainability in low-resource environments, but should not be adopted as practice without prospective multi-specimen validation.

Second, our data are consistent with current ASCO/CAP guidelines [4,5] and underscore the critical importance of minimizing cold ischemia time. Current best practice guidelines recommend fixation within 1 hour [4,5], and our results are consistent with this recommendation. Practical strategies to minimize cold ischemia time in resource-limited settings — all of which are already supported by existing ASCO/CAP guidelines rather than novel recommendations arising from our data — include: providing fixative containers directly in operating rooms; training surgical staff to place specimens immediately into fixative; establishing efficient specimen transport systems; implementing quality assurance programs tracking cold ischemia times; and including cold ischemia time information on pathology reports to aid result interpretation. These interventions require minimal financial investment but significant attention to workflow optimization and staff education.

Study limitations should be carefully considered when interpreting these results. Most importantly, we studied a single tumor specimen with strong baseline hormone receptor expression (ER 100%, PR 95–100%) and high Ki67 (40%). This single-specimen design, while standard in controlled pre-analytical studies and advantageous for eliminating inter-tumor variability as a confounding factor [8,17], inherently limits the generalizability of our findings and raises concerns of pseudoreplication: the 50 microcores, although fixed and processed independently, derive from a single biological donor and therefore do not provide independent estimates of population-level effects. The p-values reported should be read as measures of within-specimen technical variation under controlled conditions, not as licenses for population-level inference. The index tumor exhibited strong, diffuse ER and PR positivity and high Ki67, placing it far from the diagnostic decision thresholds (1% for hormone-receptor positivity, with the 1–10% “ER-low-positive” range now defined in the ASCO/CAP 2020 update [4]; 10% and 30% for Ki67 proliferation categories). In this setting, even biologically meaningful declines in staining are unlikely to shift categorical classification. Tumors with lower or heterogeneous baseline expression — particularly those with weak (1–10%) ER or PR positivity, focal or mosaic staining patterns, or Ki67 near the 10–30% range — are intrinsically closer to these thresholds and may therefore be reclassified by the same absolute magnitude of pre-analytical loss observed here. Intra-tumoral heterogeneity of ER, PR, and Ki67 is well documented in invasive breast carcinoma and was not systematically addressed by our sampling strategy; the 50 microcores were drawn from different regions of the tumor mass but may nevertheless have been enriched for regions with uniformly strong expression. Tumors with lower or moderate baseline expression, which are closer to clinically relevant diagnostic thresholds, may show greater sensitivity to pre-analytical variables and could yield different results; therefore, our observations should not be extrapolated to all breast cancer subtypes without further validation. Multi-patient studies incorporating tumors with diverse molecular subtypes, varying baseline expression levels (particularly tumors with borderline positivity), and different histological grades are needed to confirm and extend these findings before any practice recommendations can be made. The study was performed at room temperature (~25°C) typical for tropical climates; cold ischemia effects may differ at different ambient temperatures. We assessed only three biomarkers (ER, PR, and Ki67); other markers, including HER2, may show different sensitivities. Our assessment was limited to immunohistochemistry; effects on other analytical methods such as fluorescence in situ hybridization or molecular assays may differ. While we demonstrate technical impact on biomarker staining, we did not assess ultimate clinical impact on patient treatment decisions and outcomes.

We also note that representative microscopic images of the immunohistochemical staining cannot be provided for the present study. The immunohistochemical staining for ER, PR, and Ki67 on the 50 microcores was performed in 2023 at a commercial pathology laboratory operating an automated BenchMark ULTRA platform, under a fee-for-service arrangement common in resource-limited settings where in-house automated IHC is not available. The original stained slides were not archived by the commercial provider beyond their routine retention period, and digital images were not captured at the time of original assessment because the study design relied on direct microscopic scoring by two independent board-certified observers using standardized criteria. The complete quantitative scoring data for each of the 50 microcores, together with the excellent inter-observer agreement (intraclass correlation coefficient = 0.94 for ER, 0.92 for PR, and 0.96 for Ki67), are provided in the S1 Data file. We acknowledge this limitation in the methodological documentation of the present study and recommend that future studies of pre-analytical variation in resource-limited settings incorporate systematic digital image capture as part of the experimental protocol to enable visual reproducibility.

Future research should include multi-center, large-scale multi-specimen studies with diverse tumor types and varying baseline biomarker expression levels to define the relationship between initial biomarker intensity and vulnerability to cold ischemia-induced degradation. Investigation of interventions to mitigate cold ischemia effects (e.g., refrigerated temperatures, specialized transport media) would be valuable. Development of quality control markers indicating adequate pre-analytical handling would be useful. Implementation research is needed to identify effective, sustainable strategies for improving pre-analytical quality in resource-limited settings.