Comparative Efficacy of Intranasal, Intramuscular, and Intravenous Vitamin B12 Therapy for Hematological Recovery in Vitamin B12 Deficiency Anemia: A Randomized Controlled Trial

ABSTRACT

Vitamin B12 deficiency causes megaloblastic anemia and ineffective hematopoiesis. While intramuscular administration remains standard, intravenous and intranasal alternatives are increasingly used. Rigorous comparative data on hematological efficacy across routes remain limited. We thus aimed to compare intranasal, intramuscular, and intravenous vitamin B12 therapy for hematological recovery in vitamin B12 deficiency anemia, and to identify independent predictors of treatment response. In this prospective randomized controlled trial (IEC/2024/450; CTRI/2024/10/075521), 153 adults with vitamin B12 deficiency were randomized to intranasal (n = 55), intravenous (n = 56), or intramuscular (n = 42) administration. Hemoglobin (Hb), mean corpuscular volume (MCV), reticulocyte count, and lactate dehydrogenase (LDH) were assessed at baseline and on Days 7, 14, and 28. Inter‐group comparisons used the Kruskal–Wallis test; covariate‐adjusted analyzes employed ANCOVA and multiple linear regression. All three routes produced significant hematological recovery (all within‐group p < 0.001). Mean ΔHb was +2.48, +1.80, and +2.18 g/dL for intranasal, intravenous, and intramuscular groups. Although the unadjusted analysis detected an intergroup difference (H = 7.23, p = 0.027), ANCOVA abolished the significance (all p > 0.06). MCV reduction was equivalent (Kruskal–Wallis p = 0.948). LDH normalization was greater with parenteral routes (H = 8.17, p = 0.017); intramuscular therapy produced the most pronounced reticulocytosis (H = 13.93, p = 0.001). In multivariate regression (adj. R ² = 0.575), baseline Hb (β = −0.298, p < 0.001), vegan diet (β = +1.30, p = 0.001), and age (β = −0.013/year, p = 0.039) were independent predictors; route was not. All three routes produce hemoglobin recovery at 28 days with no statistically significant adjusted differences. Intranasal vitamin B12 is a safe, effective, noninvasive alternative suitable for long‐term maintenance and resource‐limited settings. All conclusions are restricted to short‐term hematological outcomes.

Trial Registration: CTRI/2024/10/075521.

1. Introduction

Vitamin B12 (cobalamin) is an indispensable water‐soluble micronutrient required for one‐carbon metabolism, DNA synthesis, myelination, and normal erythropoiesis [1]. Its deficiency results in impaired nuclear maturation, producing the characteristic megaloblastic transformation of hematopoietic precursors and, in advanced cases, pancytopenia accompanied by elevation of intracellular LDH from intramedullary hemolysis [2, 3]. Neurological manifestations, including subacute combined degeneration of the spinal cord and peripheral neuropathy, may occur independently of hematological abnormalities and can be irreversible if treatment is delayed [4].

Vitamin B12 deficiency affects an estimated 2.5%–26% of the general population, with considerably higher prevalence in South Asian populations owing to the predominance of vegetarian diets [5, 6]. In India, dietary insufficiency related to lacto‐vegetarian or vegan food practices is the most common etiology, distinguishing the population from Western cohorts where malabsorption secondary to pernicious anemia or gastrointestinal surgery predominates [6, 7].

Intramuscular (IM) cyanocobalamin or hydroxocobalamin has historically been the standard replacement therapy, particularly in malabsorption states [8]. Intravenous (IV) administration achieves rapid correction and is employed when IM access is impractical or when the clinical presentation demands prompt normalization. Intranasal (IN) vitamin B12 exploits the rich vascularity of the nasal mucosa to deliver cobalamin systemically while bypassing the gastrointestinal tract, and has demonstrated adequate bioavailability in several pharmacokinetic studies [9, 10]. Its noninvasive nature makes it an attractive long‐term maintenance option.

Oral high‐dose vitamin B12 supplementation has also emerged as a viable noninvasive alternative. Randomized trials have demonstrated that cyanocobalamin 1000–2000 μg daily achieves hematological and biochemical recovery comparable to intramuscular therapy, leveraging passive intestinal diffusion at supraphysiological doses independent of intrinsic factor [11]. Several international guidelines now endorse oral therapy as a first‐line option for most patients without severe neurological compromise. Despite the availability of these routes, direct head‐to‐head comparative randomized data on hematological efficacy are sparse. Most published work evaluates individual routes in isolation, and systematic reviews have highlighted the paucity of adequately powered comparative trials with covariate‐adjusted analysis [12]. Importantly, prior studies have not employed covariate‐adjusted analyzes to disentangle the influence of baseline disease severity and dietary etiology on apparent treatment differences—a potentially significant source of confounding given the heterogeneity of presentations in clinical practice.

The present randomized controlled trial was therefore designed to compare the hematological response to intranasal, intramuscular, and intravenous vitamin B12 therapy in a cohort of adults with confirmed vitamin B12 deficiency anemia presenting to a tertiary care center, and to identify independent determinants of recovery magnitude through multiple linear regression.

2. Methods

2.1. Study Design and Setting

This was a prospective, open‐label, three‐arm randomized controlled trial conducted at a tertiary care teaching hospital over a 12‐month period following approval by the Institutional Ethics Committee (IEC No. IEC/2024/450, dated February 7, 2024). The trial was prospectively registered with the Clinical Trials Registry of India (CTRI/2024/10/075521). The study was conducted in accordance with the Declaration of Helsinki. Written informed consent was obtained from all participants before enrolment.

2.2. Study Population

Adults (≥ 18 years) presenting with laboratory‐confirmed vitamin B12 deficiency (serum B12 < 200 pg/mL) accompanied by hematological abnormalities consistent with megaloblastic anemia were screened for eligibility. Patients were excluded if they had coexisting folate deficiency, iron deficiency anemia, pregnancy, anemia requiring immediate blood transfusion, chronic kidney disease, or hematological malignancy to ensure that observed hematological changes were attributable solely to B12 repletion (Consort diagram—Figure 1).

FIGURE 1.

CONSORT flow diagram showing patient enrolment, allocation, follow‐up, and analysis. [Color figure can be viewed at wileyonlinelibrary.com]

2.3. Sample Size

The sample size was calculated to detect a minimum clinically significant difference of 0.75 g/dL in hemoglobin between groups, assuming a pooled standard deviation of 1.5 g/dL, with two‐sided α = 0.05 and 80% power. This yielded a minimum requirement of 40 patients per group. To account for expected attrition and ensure adequate power for secondary analyzes, 153 patients were enrolled. Post hoc power analysis confirmed that the study had > 85% power to detect the specified effect size under the three‐group repeated‐measures design.

2.4. Randomization and Intervention

Eligible participants were randomized in a 1:1:1 ratio using a simple randomization procedure (chit system), with allocation concealment maintained via sequentially numbered, sealed, opaque envelopes prepared by a biostatistician independent of the clinical team prior to trial initiation; envelopes were stored in the pharmacy and accessed only at the point of allocation. Stratified randomization according to anemia severity or other baseline characteristics was not employed, as this was considered operationally impractical in the three‐arm clinical setting; this is acknowledged as a limitation contributing to observed baseline imbalances. The three treatment arms were:

Intranasal group (n = 55): Vitamin B12 (methylcobalamin) 250 μg per spray administered into each nostril on alternate days for seven doses (Days 1–13), followed by once‐weekly dosing for a further 4 weeks.

Intramuscular group (n = 42): Cyanocobalamin 1000 μg injected into the gluteal muscle on alternate days for seven doses (Days 1–13), followed by once‐weekly administration for 4 weeks.

Intravenous group (n = 56): Cyanocobalamin 1000 μg diluted in 100 mL normal saline infused intravenously over 30 min on alternate days for seven doses (Days 1–13), followed by once‐weekly administration for 4 weeks.

2.5. Outcome Measures

The primary outcome was the change in hemoglobin concentration (ΔHb) from baseline to Day 28. Secondary outcomes included change in MCV, total leucocyte count (TLC), platelet count, reticulocyte percentage, and serum LDH. All parameters were assessed at baseline and on Days 7, 14, and 28. Resolution of hypersegmented neutrophils on peripheral blood smear was also recorded as a qualitative hematological endpoint.

2.6. Statistical Analysis

Continuous variables are presented as mean ± standard deviation. Normality of change scores was assessed using the Shapiro–Wilk test. Where nonnormality was confirmed (W < 0.95, p < 0.05), nonparametric methods were used for inter‐group comparisons. Within‐group change from baseline to each follow‐up time point was assessed by a paired t‐test. Inter‐group comparisons of change scores used the Kruskal–Wallis test, with pairwise Mann–Whitney U tests for post hoc analysis.

To address baseline imbalance in hemoglobin and LDH between groups, covariate‐adjusted analysis of covariance (ANCOVA) was performed with baseline Hb, age, sex, baseline B12, and baseline MCV as covariates. Multiple linear regression models were constructed for ΔHb, ΔMCV, and ΔLDH at Day 28, incorporating route of administration, age, sex, baseline laboratory values, alcohol use, smoking status, and dietary category (vegetarian, vegan, ovo‐lacto, nonvegetarian; reference: vegetarian) as predictors. Model fit was assessed by adjusted R ² and the overall F‐statistic. A p‐value < 0.05 was considered statistically significant throughout. Fifteen patients (all intramuscular group) had missing Day‐28 data due to administrative reasons unrelated to the intervention and were excluded from multivariable regression analyzes (n = 138); missing data were not distributed across groups, representing a potential source of attrition bias. A linear mixed‐effects repeated‐measures model (LMEM) is recommended as a sensitivity approach in future studies. Fifteen patients, all from the intramuscular group, had missing Day‐28 outcome data due to administrative reasons unrelated to the intervention and were excluded from multivariable regression analyzes (analytic sample n = 138); missing data were not distributed across groups, representing a potential source of attrition bias. A linear mixed‐effects repeated‐measures model (LMEM) would represent an appropriate sensitivity approach for longitudinal trajectory analysis and is recommended for future studies.

3. Results

3.1. Baseline Characteristics

A total of 153 patients were enrolled and randomized: intranasal n = 55, intravenous n = 56, intramuscular n = 42. The overall cohort had a mean age of 39.1 ± 14.4 years (range 18–82); 81 (52.9%) were male. Dietary patterns comprised vegetarian in 85 (55.6%), nonvegetarian in 41 (26.8%), ovo‐lacto in 17 (11.1%), and strict vegan in 10 (6.5%). The most common modes of clinical presentation were asymptomatic detection/incidental finding (28.8%), bicytopenia (17.6%), pancytopenia (14.4%), and fatiguability (10.5%). Baseline hematological and demographic parameters are summarized in Table 1.

TABLE 1.

Baseline demographic and hematological characteristics by treatment group.

Parameter	Intranasal (n = 55)	Intravenous (n = 56)	Intramuscular (n = 42)
Age (years)	36.7 ± 14.7	38.0 ± 13.3	43.6 ± 14.7
Male sex, n (%)	29 (52.7%)	30 (53.6%)	22 (52.4%)
Serum B12 (pg/mL)	76.9 ± 44.5	83.8 ± 42.1	92.8 ± 36.9
Hemoglobin (g/dL)	8.73 ± 3.34	9.99 ± 3.43	10.25 ± 2.19
MCV (fL)	108.8 ± 13.0	109.1 ± 10.7	106.5 ± 10.4
LDH (IU/L)	548.8 ± 570.1	1255.8 ± 1218.2	1230.0 ± 1070.7
Reticulocytes (%)	0.68 ± 0.43	0.73 ± 0.28	0.76 ± 0.36

Note: Values are mean ± SD unless otherwise stated.

Abbreviations: LDH = lactate dehydrogenase; MCV = mean corpuscular volume.

Notably, the intravenous and intramuscular groups had substantially higher baseline LDH levels (~1230–1256 IU/L) compared to the intranasal group (~549 IU/L), reflecting greater baseline ineffective erythropoiesis in the parenteral groups. Conversely, the intranasal group had a lower mean baseline Hb (8.73 g/dL) compared to the intravenous (9.99 g/dL) and intramuscular (10.25 g/dL) groups. These imbalances, expected under simple randomization in a relatively small trial, were addressed in subsequent covariate‐adjusted analyzes.

3.2. Hemoglobin Response

All three treatment groups demonstrated a progressive and statistically significant rise in hemoglobin over the 28‐day study period. Mean Hb values across all timepoints are presented in Table 2. By Day 28, mean Hb had increased by +2.48 g/dL in the intranasal group (paired t = −10.89, p < 0.001), +2.18 g/dL in the intramuscular group (paired t = −10.06, p < 0.001), and +1.80 g/dL in the intravenous group (paired t = −9.24, p < 0.001).

TABLE 2.

Mean hemoglobin (g/dL) by treatment group and timepoint.

Timepoint	Intranasal	Intravenous	Intramuscular	p a
Baseline	8.73 ± 3.34	9.99 ± 3.43	10.25 ± 2.19	—
Day 7	9.48 ± 3.22	10.46 ± 3.12	10.85 ± 2.07	0.081
Day 14	10.32 ± 3.04	11.06 ± 3.02	11.56 ± 1.89	0.052
Day 28	11.21 ± 2.78	11.79 ± 2.96	12.43 ± 2.04	0.027 b
ΔHb (Day 28)	+2.48 ± 1.69	+1.80 ± 1.46	+2.18 ± 1.15	0.027 b

Note: Values are mean ± SD.

^a Kruskal–Wallis test for ΔHb.

^b Unadjusted; intergroup significance abolished on ANCOVA adjustment.

An unadjusted intergroup comparison using the Kruskal–Wallis test detected a statistically significant overall difference in ΔHb at Day 28 (H = 7.23, p = 0.027). Post hoc Mann–Whitney analysis identified the source of this difference as the intranasal versus intravenous comparison (U = 1969, p = 0.012), while intranasal versus intramuscular (p = 0.762) and intravenous versus intramuscular (p = 0.067) were not significantly different. Shapiro–Wilk testing confirmed nonnormality of ΔHb in the intranasal (W = 0.935, p = 0.005) and intravenous (W = 0.913, p = 0.001) groups, supporting the use of nonparametric methods for inter‐group comparison.

However, this unadjusted difference was attributable to the substantially lower baseline Hb in the intranasal group. ANCOVA controlling for baseline Hb, age, sex, baseline B12, and baseline MCV showed no statistically significant difference between any pair of treatment groups in absolute Hb at Day 28 (intranasal vs. intravenous: β = +0.256, p = 0.22; intramuscular vs. intravenous: β = +0.468, p = 0.067; adjusted R ² = 0.801).

3.3. Mean Corpuscular Volume

MCV declined progressively in all groups throughout the follow‐up period, reflecting resolution of megaloblastic erythropoiesis. By Day 28, mean MCV reductions were −10.33 ± 10.42 fL (intranasal; paired t = 7.35, p < 0.001), −10.14 ± 6.05 fL (intravenous; paired t = 12.55, p < 0.001), and −9.68 ± 8.67 fL (intramuscular; paired t = 5.91, p < 0.001). The inter‐group comparison of ΔMCV by the Kruskal–Wallis test showed no significant difference (H = 0.053, p = 0.948), indicating no statistically significant between‐group difference in MCV normalization across the three routes. MCV data are presented in Table 3.

TABLE 3.

Mean MCV (fL) by treatment group and timepoint.

Timepoint	Intranasal	Intravenous	Intramuscular	p a
Baseline	108.8 ± 13.0	109.1 ± 10.7	106.5 ± 10.4	—
Day 7	104.8 ± 11.6	105.9 ± 9.8	103.5 ± 9.2	0.614
Day 14	102.0 ± 9.8	102.4 ± 8.6	100.1 ± 8.4	0.511
Day 28	98.5 ± 8.3	99.0 ± 7.9	96.8 ± 8.0	0.340
ΔMCV (Day 28)	−10.33 ± 10.42	−10.14 ± 6.05	−9.68 ± 8.67	0.948

Note: Values are mean ± SD.

^a Kruskal–Wallis test; all within‐group changes p < 0.001 by paired t‐test.

3.4. Lactate Dehydrogenase

LDH declined significantly over time in all groups. The magnitude of LDH reduction from baseline at Day 28 was −300.5 ± 502.1 IU/L (intranasal), −868.7 ± 1009.4 IU/L (intravenous), and −637.1 ± 651.4 IU/L (intramuscular). Kruskal–Wallis analysis of ΔLDH at Day 28 was statistically significant (H = 8.17, p = 0.017). Post hoc pairwise analysis showed that both the intravenous (U = 1967, p = 0.012) and intramuscular (U = 1006, p = 0.023) groups achieved greater LDH reduction than the intranasal group, while the two parenteral routes were not significantly different from each other (p = 0.769). The absolute Day 28 LDH values converged across groups (intranasal 248.3, intravenous 387.1, intramuscular 592.9 IU/L), reflecting the substantially higher baseline in the parenteral groups. LDH data are presented in Table 4.

TABLE 4.

Mean LDH (IU/L) by treatment group and timepoint.

Timepoint	Intranasal	Intravenous	Intramuscular	p a
Baseline	548.8 ± 570.1	1255.8 ± 1218.2	1230.0 ± 1070.7	< 0.001
Day 7	413.7 ± 498.2	948.0 ± 892.4	852.5 ± 768.3	0.004
Day 14	317.8 ± 420.6	658.3 ± 712.2	350.2 ± 490.1	0.009
Day 28	248.3 ± 392.1	387.1 ± 521.8	592.9 ± 602.4	0.017
ΔLDH (Day 28)	−300.5 ± 502.1	−868.7 ± 1009.4	−637.1 ± 651.4	0.017 b

Note: Values are mean ± SD.

^a Kruskal–Wallis test for ΔLDH.

^b IV and IM both significantly greater than IN on post hoc analysis (p = 0.012 and p = 0.023, respectively).

3.5. Reticulocyte Response

An early reticulocyte response was observed in all groups, consistent with effective restoration of erythropoiesis. Kruskal–Wallis analysis of the change in reticulocyte count at Day 7 revealed a highly significant intergroup difference (H = 13.93, p = 0.001). The intramuscular group showed the most marked reticulocytosis, with a mean ΔRetic of +0.74% at Day 7 and a peak response of +2.57% at Day 14—a pattern characteristic of the reticulocyte crisis typically observed with parenteral cyanocobalamin therapy [13]. Reticulocyte changes in the intranasal (+0.28% Day 7, +0.58% Day 14) and intravenous (+0.48% Day 7, +0.90% Day 14) groups were more gradual but still reflected progressive marrow recovery.

3.6. Peripheral Blood Smear

Hypersegmented neutrophils, a hallmark of megaloblastic change, were present at baseline in 26 of 55 intranasal patients (47.3%), 31 of 56 intravenous patients (55.4%), and 22 of 42 intramuscular patients (52.4%). Complete resolution was documented in all patients with available Day 28 peripheral smear data across all three treatment groups, confirming effective eradication of the megaloblastic state irrespective of route.

3.7. Multivariate Analysis of Predictors of Treatment Response

Multiple linear regression was performed for ΔHb, ΔMCV, and ΔLDH at Day 28 to identify independent predictors of treatment response and to adjust for the baseline imbalances inherent in the randomization. Full model results are presented in Table 5 (Figures 2, 3).

TABLE 5.

Multiple linear regression: Independent predictors of hematological change at Day 28.

Predictor	Outcome	β	SE	t	p	Sig.
Baseline Hb (per g/dL)	ΔHb	−0.298	0.033	−9.01	< 0.001	***
Vegan versus vegetarian diet	ΔHb	+1.296	0.374	3.47	0.001	***
Ovo‐lacto versus vegetarian diet	ΔHb	+0.958	0.298	3.22	0.002	**
Age (per year)	ΔHb	−0.013	0.006	−2.09	0.039	*
Baseline MCV (per fL)	ΔHb	+0.017	0.008	2.02	0.045	*
Admin: Intranasal vs. IV a	ΔHb	+0.214	0.210	1.02	0.311	ns
Admin: IM versus IV a	ΔHb	+0.338	0.245	1.38	0.171	ns
Baseline MCV (per fL)	ΔMCV	−0.590	0.035	−16.6	< 0.001	***
Male sex	ΔMCV	−2.320	0.842	−2.76	0.007	**
Admin: IM versus IV a	ΔMCV	−2.342	1.046	−2.24	0.027	*
Baseline LDH (per IU/L)	ΔLDH	−0.847	0.014	−58.9	< 0.001	***
Admin: IM versus IV a	ΔLDH	−82.7	32.4	−2.55	0.012	*

Note: *p < 0.05; **p < 0.01; ***p < 0.001; ns = not significant.

Abbreviations: IM = intramuscular; IV = intravenous.

^a Reference category: Intravenous. n = 138 in all models. Model adj. R ²: ΔHb = 0.575; ΔMCV = 0.755; ΔLDH = 0.974.

FIGURE 2.

(A) Adjusted R ² for all three regression models. (B) ANCOVA‐adjusted Hb at Day 28 by route (error bars = 95% CI; no significant intergroup difference). (C) Unadjusted ΔHb trajectories (*p = 0.012 unadjusted; NS on ANCOVA). [Color figure can be viewed at wileyonlinelibrary.com]

FIGURE 3.

Forest plot of multivariate regression coefficients for Models 1 (ΔHb), 2 (ΔMCV), and 3 (ΔLDH). Filled squares = p < 0.05; open circles = p ≥ 0.05. Error bars = 95% CI. [Color figure can be viewed at wileyonlinelibrary.com]

3.7.1. Predictors of ΔHb (Model 1; n  = 138; adjusted R ² = 0.575; F  = 15.24, p  < 0.001)

Baseline Hb was the single strongest independent predictor of ΔHb (β = −0.298, SE = 0.033, p < 0.001), confirming a regression‐to‐the‐mean effect whereby patients with more severe anemia at baseline showed greater absolute improvement (Figure 4). Dietary pattern was a significant independent predictor after adjustment: vegans gained an additional +1.30 g/dL (β = +1.296, p = 0.001) compared to vegetarians, plausibly reflecting near‐total dietary cobalamin exclusion and consequent greater tissue depletion. The association in ovo‐lacto patients (+0.96 g/dL, β = +0.958, p = 0.002) requires more cautious interpretation: eggs and dairy contain bioavailable cobalamin, and the argument of greater dietary restriction does not apply uniformly to this group. This finding may reflect variability in consumption frequency, inter‐individual differences in bioavailability from these sources, or residual unmeasured confounding, and should be regarded as hypothesis‐generating pending independent replication in larger, diet‐stratified cohorts. Age was a modest but significant negative predictor (β = −0.013 g/dL per year, p = 0.039), consistent with declining bone marrow reserve in older patients [14]. Critically, after adjustment for these covariates, route of administration (intranasal vs. intravenous β = +0.214, p = 0.311; intramuscular vs. intravenous β = +0.338, p = 0.171) was not an independent predictor of ΔHb, indicating that route of administration was not independently associated with hemoglobin recovery in the covariate‐adjusted model.

FIGURE 4.

Key independent predictors of ΔHb. (A) Adjusted β‐coefficients for dietary pattern (vs. vegetarian reference). (B) Scatter of baseline Hb versus ΔHb at Day 28 by route, with regression line (β = −0.298, p < 0.001). [Color figure can be viewed at wileyonlinelibrary.com]

3.7.2. Predictors of ΔMCV (Model 2; n  = 138; adjusted R ² = 0.755; F  = 33.44, p < 0.001)

Baseline MCV was the dominant predictor of MCV normalization (β = −0.590, p < 0.001), indicating that for each 1‐unit increase in baseline MCV, 59% of any elevation above normal is reversed. Male sex was an independent predictor of greater MCV reduction (β = −2.32 fL, p = 0.007), possibly reflecting higher erythropoietic turnover in male patients. The intramuscular route was associated with a significantly greater MCV reduction compared with intravenous (β = −2.34 fL, p = 0.027), suggesting that the more pronounced reticulocyte crisis and earlier marrow response with IM therapy translate into modestly faster normalization of red cell indices.

3.7.3. Predictors of ΔLDH (Model 3; n  = 138; adjusted R ² = 0.974; F  = 397.4, p < 0.001)

Baseline LDH overwhelmingly dominated this model (β = −0.847, p < 0.001), indicating that approximately 84.7% of baseline LDH elevation is resolved by Day 28, irrespective of route, age, sex, or diet. This near‐unity coefficient (β = −0.847) is consistent with LDH as a quantitatively predictable treatment‐response biomarker in megaloblastic anemia. It should be noted, however, that the high adjusted R ² of 0.974 partly reflects a mathematical dependency when a change score is regressed on its own baseline and should be interpreted in this context [15]. After adjustment for baseline LDH, the intramuscular route showed a residual additional LDH reduction of −82.7 IU/L compared to intravenous (β = −82.7, p = 0.012), suggesting a modest but statistically independent suppression of ineffective erythropoiesis beyond what baseline severity alone predicts (Figure 5).

FIGURE 5.

LDH normalization trajectories (A) and reticulocyte response (B) by treatment route. KW p = 0.017 for ΔLDH at Day 28; KW p = 0.001 for reticulocyte change at Day 7. [Color figure can be viewed at wileyonlinelibrary.com]

4. Discussion

This randomized controlled trial comparing three routes of vitamin B12 replacement in patients with megaloblastic anemia yields three principal insights: (i) all three routes produce clinically meaningful and statistically significant hematological recovery at 28 days, with no statistically significant adjusted differences between groups; (ii) apparent superiority of the intranasal route in unadjusted Hb analyzes is an artifact of greater baseline anemia severity in that group, disappearing on covariate adjustment; and (iii) route of administration is associated with differences in the kinetics of erythropoietic recovery as captured by LDH and reticulocyte dynamics, with intramuscular therapy showing the most rapid early response.

The absence of statistically significant adjusted differences between routes in Hb recovery at 4 weeks is consistent with existing evidence. The cornerstone randomized study by Slot et al. demonstrated effective systemic bioavailability of intranasal cobalamin [9], and a subsequent systematic review and network meta‐analysis by Abdelwahab and colleagues concluded that no single route of B12 supplementation demonstrated statistically significant superiority in biochemical or hematological outcomes when adequate dosing was maintained [12]. The current study extends these findings with a prospective randomized design and the additional rigor of covariate‐adjusted analysis. Our ANCOVA model (adjusted R ² = 0.801) demonstrated that, after controlling for baseline Hb, age, sex, and baseline MCV, neither intranasal (β = +0.256, p = 0.22) nor intramuscular (β = +0.468, p = 0.067) therapy produced a significantly different absolute Hb at Day 28 compared to intravenous therapy.

A particularly novel finding of this study is the independent and clinically substantial effect of dietary pattern on the magnitude of Hb recovery. After full covariate adjustment, vegan patients gained an additional 1.30 g/dL and ovo‐lacto patients 0.96 g/dL compared to vegetarians—a difference exceeding the prespecified minimum clinically significant threshold of 0.75 g/dL. This is unlikely to reflect differential bioavailability across routes; instead, it suggests that patients with more restrictive dietary B12 exclusion present with more prolonged and complete tissue B12 depletion, thereby providing greater hematopoietic reserve for recovery once repletion is initiated. Importantly, serum B12 level at presentation did not independently predict ΔHb (β = +0.001, p = 0.664), underscoring the well‐established limitation of a single serum B12 measurement as an index of total body B12 stores [1, 3]. The dietary signal captured in this regression model likely serves as a more reliable proxy for the chronicity of deficiency.

The significant intergroup difference in LDH normalization (Kruskal–Wallis p = 0.017), with both parenteral routes achieving greater reductions than intranasal therapy, may reflect the more rapid and complete elevation of serum cobalamin concentrations achieved by intravenous and intramuscular administration, allowing earlier suppression of intramedullary hemolysis [16]. The LDH regression model (adjusted R ² = 0.974) demonstrates that approximately 84.7% of baseline LDH elevation resolves by Day 28 irrespective of route, consistent with the known reversibility of megaloblastic intramedullary haemolysis with adequate cobalamin repletion. It should be acknowledged that the high R ² in part, reflects a mathematical dependency when a change score is regressed on its own baseline; nevertheless, the near‐unity coefficient (β = −0.847) confirms the highly predictable and quantifiable nature of LDH normalization as a treatment‐response marker.

The more pronounced reticulocyte response in the intramuscular group (peak ΔRetic +2.57% at Day 14; Kruskal–Wallis at Day 7: H = 13.93, p = 0.001) is consistent with the kinetics of the “reticulocyte crisis” classically described after parenteral cyanocobalamin injection, typically peaking between Days 5 and 10 after initiation of therapy [13]. The intramuscular route delivers a depot of cobalamin into well‐vascularised muscle tissue with a more sustained release profile than intravenous bolus dosing. An additional pharmacokinetic consideration is that intravenous administration achieves supraphysiological peak serum cobalamin concentrations that exceed the renal transport threshold, resulting in substantial urinary excretion of the administered dose within hours of infusion. Intramuscular delivery, by sustaining erythropoietically relevant tissue concentrations over a longer period, may thereby provide more effective early marrow stimulation. Although urinary B12 was not measured in this study, this mechanism represents a plausible pharmacokinetic explanation for the more pronounced reticulocyte crisis and the modestly superior MCV normalization observed in the regression model (β = −2.34 fL, p = 0.027). Future trials should incorporate urinary B12 measurement to test this hypothesis directly.

The age‐related attenuation of Hb recovery (−0.013 g/dL per year, p = 0.039) identified in the regression analysis is consistent with the known age‐dependent decline in hematopoietic stem cell function and erythropoietic reserve and has practical implications for the expected treatment response in older patients with B12 deficiency anemia [14].

From a clinical practice perspective, the absence of significant adjusted differences in hemoglobin recovery between routes has important implications. It offers a needle‐free, self‐administered alternative suitable for community and outpatient settings, and may enhance adherence in patients requiring prolonged maintenance therapy, a group that includes those with pernicious anemia, gastrointestinal malabsorption, and those unwilling to undergo repeated parenteral injections [11]. These advantages are particularly relevant in resource‐limited settings and in pediatric or elderly populations where repeated intramuscular injections may be poorly tolerated.

4.1. Limitations

This study has several limitations that merit acknowledgement. First, simple randomization produced baseline imbalances in Hb and LDH requiring post hoc adjustment; stratified or minimisation‐based randomization is recommended for future three‐arm trials. Second, 15 patients (all intramuscular arm) had missing Day‐28 data; complete‐case analyzes (n = 138) may introduce attrition bias given nonrandom distribution of missing data. Third, reticulocyte sampling at Days 7, 14, and 28 did not capture the peak reticulocyte crisis (Days 5–10); future studies should include Days 3 and 5.

Fourth, follow‐up was limited to 28 days. Longer‐term data would inform the durability of recovery. Fifth, the predominantly dietary etiology limits generalisability to malabsorption‐predominant populations (e.g., pernicious anemia, postgastrectomy). Sixth, blinding was not feasible; laboratory analyzes were performed by personnel unaware of group allocation. Seventh, all conclusions are explicitly restricted to short‐term hematological outcomes; neurological and hematological responses do not always parallel each other, and route‐dependent differences in neural tissue B12 delivery require prospective evaluation. Eighth, a linear mixed‐effects repeated‐measures model is recommended as a sensitivity approach. Ninth, urinary B12 measurement in future trials would quantify renal losses after intravenous dosing.

5. Conclusion

In patients with vitamin B12 deficiency anemia of predominantly dietary etiology, intranasal, intramuscular, and intravenous routes of replacement therapy produce hemoglobin recovery and MCV normalization at 28 days with no statistically significant adjusted differences between routes. The apparent advantage of intranasal therapy in unadjusted analyzes is attributable to greater baseline anemia severity rather than superior therapeutic efficacy. Intramuscular administration achieves more rapid and pronounced suppression of ineffective erythropoiesis, as evidenced by a more vigorous reticulocyte response and significantly greater LDH reduction, and produces modestly faster MCV normalization. The dominant independent predictors of hematological recovery are baseline hemoglobin severity, age, and dietary pattern (vegan diet a significant independent predictor; the ovo‐lacto association is hypothesis‐generating and requires independent replication).

These findings support intranasal vitamin B12 as a clinically effective, non‐invasive alternative to parenteral therapy for patients with dietary B12‐deficiency anemia without significant malabsorption. All conclusions are restricted to short‐term hematological outcomes; neurological comparability requires prospective evaluation. For patients with severe disease or where rapid erythropoietic recovery is clinically prioritized, intramuscular therapy may be preferred. Future trials with longer follow‐up, stratified randomization, and neurological outcome assessment are warranted to further define the optimal route‐selection strategy in diverse clinical contexts.

Funding

The authors have nothing to report.

Ethics Statement

Institutional Ethics Committee, IEC/2024/450, February 7, 2024.

Consent

Written informed consent was obtained from all participants prior to enrolment in the study. Participants were informed of the study objectives, procedures, potential risks, and their right to withdraw at any time without consequence to their clinical care. The consent process was conducted in accordance with the Declaration of Helsinki and the requirements of the Institutional Ethics Committee (IEC/2024/450). Patient data were fully anonymized; no identifying information is included in this manuscript.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.