Prevalence and Associations of Abnormal Levels of Vitamin B6 in People with Parkinson's Disease Authors: Karlo J. Lizarraga, Lu Wang, Hannah Mesmer, Nami Shah, William Barbosa, Eric Logigian, Michael P. McDermott, Anthony E. Lang Published: 2026 ABSTRACT AbstractBackgroundIn people with Parkinson's disease, abnormal levels of vitamin B6 may be associated with polyneuropathy and epileptic seizures, including medication‐refractory status epilepticus.ObjectivesTo estimate the prevalence and explore associations of abnormal levels of vitamin B6 in people with Parkinson's disease.MethodsIn this cross‐sectional study, fasting serum levels of vitamin B6 and related metabolites were measured in 144 people with Parkinson's disease randomly recruited from eligible clinic patients, compared with 144 historical controls matched for age, sex, diet, and vitamin supplementation. Mixed‐effects logistic regression was used to compare the prevalences of abnormal B6 levels, and mixed‐effects analysis of variance was used to compare mean B6 levels between groups. Mixed effects regression models were used to examine associations between B6 levels and pre‐defined demographic and clinical variables.ResultsMean age was 71 years (range = 43–87) and 38.2% were female. Almost all were treated with carbidopa/levodopa (n = 140/144, mean levodopa dosage = 800 mg/day, mean treatment duration = 78 months). The prevalence of B6 deficiency in people with Parkinson's disease was 36.1% compared to 4.9% in controls (OR = 11.4, 95% CI = 4.9–26.5, p < 0.001). The average B6 level in people with Parkinson's disease was approximately half of that in controls (ratio of geometric means = 0.49, 95% CI = 0.41–0.58, p < 0.001). The severity of levodopa‐associated motor complications (p = 0.04) and the presence of epileptic seizures (p = 0.006) were associated with lower log‐transformed B6 levels.ConclusionsVitamin B6 deficiency is highly prevalent in people with Parkinson's disease. Active monitoring should be conducted in those with levodopa‐associated motor complications or epileptic seizures. The prevalence of epilepsy in people with Parkinson's disease who live in nursing homes is 8% and the prevalence of polyneuropathy is 30–40% compared to 1% and 10%, respectively, in older adults. 1 , 2 It is unclear why the prevalence of these problems is increased in people with Parkinson's disease. Pyridoxine (vitamin B6) is an essential cofactor for the synthesis and catabolism of essential neurotransmitters and amino acids. 3 Both low and high B6 levels have been associated with polyneuropathy in the general population. 4 People with Parkinson's disease might be at risk of developing abnormal B6 levels due to age, dietary habits, vitamin supplement misuse, gastrointestinal dysfunction, and complex interactions with carbidopa/levodopa. 3 , 5 B6 deficiency could lead to polyneuropathy and epilepsy, which are treatable and potentially preventable complications associated with disability and death in people with Parkinson's disease. 3 , 5 In a systematic review, we estimated a 41.4% relative frequency of abnormal B6 levels in people with Parkinson's disease. 5 Available data suggest a possible association between low B6 levels and relatively high dosages of carbidopa/levodopa, and possible associations between low B6 levels, polyneuropathy, and epilepsy, 5 , 6 including four published cases of medication‐refractory status epilepticus. 7 , 8 , 9 , 10 The prevalence and clinical importance of abnormal B6 levels in people with Parkinson's disease remain unclear as few data are available from people without polyneuropathy or epilepsy. In the U.S., the Food and Drug Administration (FDA) has recently issued a drug safety communication regarding seizures associated with low B6 levels in people with Parkinson's disease treated with carbidopa/levodopa. 11 We conducted a cross‐sectional study to estimate the prevalence and explore associations of abnormal B6 levels in people with Parkinson's disease compared to matched historical controls. We hypothesized that (1) people with Parkinson's disease have a significantly higher prevalence of B6 deficiency compared to controls, (2) B6 levels are associated with similar demographic variables in people with Parkinson's disease and controls, and (3) B6 levels are associated with disease duration and severity, levodopa dosage, polyneuropathy, and epilepsy in people with Parkinson's disease. Methods The study was approved by the Research Subjects Review Board of the University of Rochester Medical Center (URMC) on June 6, 2024 (STUDY00009245). Written informed consent was obtained from all patients participating in the study. The study is registered on ClinicalTrials.gov (https://clinicaltrials.gov/study/NCT07430722). The study protocol and statistical analysis plan are available at the University of Rochester Research Repository (https://doi.org/10.60593/ur.d.31424636). We adhered to the STROBE guidelines for reporting observational studies. Study Design and Participants This was a cross‐sectional study of people with Parkinson's disease recruited from randomly generated lists of eligible patients from the Movement Disorders Clinic of the URMC, compared with matched historical controls from the publicly available database of the U.S. National Health and Nutrition Examination Survey (NHANES). 12 , 13 Inclusion Criteria People older than 18 years with Parkinson's disease according to the Movement Disorders Society criteria, 14 who were able to speak and understand English or Spanish fluently. Exclusion Criteria People with comorbidities (renal disease, bowel disease, rheumatoid arthritis) or taking medications associated with abnormal B6 levels (steroids, cycloserine, hydralazine, isoniazid, theophylline, penicillamine) other than levodopa. Procedures and Outcomes A REDCap database was built to collect and manage data for the study. 15 Randomization and Enrollment Every Monday between July 2024 and June 2025, Microsoft Excel was used to create a list of chronological time slots assigned to all morning clinic appointments scheduled for the next week. Numeric values were randomly assigned to each time slot using the “RAND” function. A pre‐defined phone call script was used to contact potential study participants starting at the highest numeric value, followed by the second highest numeric value, and so on until pre‐enrolling 3–4 participants per week. If the potential study participant expressed interest, the informed consent form and the structured interview form were sent electronically. The study visit was scheduled for the same day as their upcoming clinic visit, and potential participants were encouraged to fast from the night prior to the study visit. Written informed consent was obtained during the study visit. Laboratory Data A trained member of the study team (HM) collected and processed blood samples following pre‐specified protocols. The samples were sent to the URMC laboratory for measurement of serum levels of vitamin B12 (normal = 232–1245 pg/mL), folic acid (normal ≥4.6 ng/mL), and homocysteine (normal ≤15 μmol/L), and to the Associated Regional and University Pathologists Laboratory for measurement of serum levels of vitamin B6 (pyridoxal 5′‐phosphate, normal = 20–125 nmol/L) and methylmalonic acid (normal ≤0.4 μmol/L). Demographic Data Structured interviews were conducted based on standardized questions of the NHANES. 12 , 13 Data were collected on age (years), sex (male/female), race, ethnicity, body mass index (kg/m2), history of diabetes (yes/no), smoking (yes/no), alcohol use (yes/no) and oral contraceptive pill use (yes/no), food insecurity (yes/no), daily consumption of protein (yes/no) and food rich in B6 (poultry, fish, greens, tropical fruits) (yes/no) in the last 30 days, and use of vitamin supplementation (yes/no) and B6 supplementation (yes/no) in the last 30 days. Interviews were conducted on the same day as the blood draw. Thus, the interviewers were blinded to the results of the laboratory tests. Clinical Data Data in people with Parkinson's disease included age at symptom onset (years), disease duration (years), Hoehn & Yahr stage, 16 levodopa dosage (mg/day), levodopa treatment duration (months), dopa decarboxylase inhibitor dosage (mg/day), carbidopa/levodopa dosage ratio, cathecol‐o‐methyl‐transferase inhibitor use (yes/no), levodopa equivalent daily dosage (mg/day), 17 treatment with deep brain stimulation (yes/no), Movement Disorders Society sponsored version of the Unified Parkinson's Disease Rating Scale (MDS‐UPDRS): MDS‐UPDRS‐I (non‐motor symptoms), MDS‐UPDRS‐II (motor symptoms), and MDS‐UPDRS‐IV (motor complications), 16 and paroxysmal episodes consistent with epileptic seizures (yes/no/uncertain). Three trained clinicians (KJL, NS, WB) conducted a neurological examination to obtain the MDS‐UPDRS‐III (motor examination) in the dopaminergic ON state, and the modified Toronto Clinical Neuropathy Score (mTCNS). 18 We used the value of ≥3 points in the mTCNS to define polyneuropathy. 19 Examinations were conducted on the same day as the blood draw. Thus, the examiners were blinded to the results of the laboratory tests. Control Data The same laboratory and demographic data were collected from the publicly available database of the NHANES 2003–2004. 12 , 13 The “RAND” function on Microsoft Excel was used to randomly choose controls matched for age (years), sex, consumption of protein (yes/no), consumption of food rich in B6 (yes/no), vitamin supplementation use (yes/no), and B6 supplementation use (yes/no). Controls with comorbidities or medication exposures associated with abnormal B6 levels, and those who were not fasting at the time of venipuncture, were excluded. Statistical Analyses Based on our systematic review, 5 we calculated that a sample size of 144 people with Parkinson's disease would be necessary to estimate the prevalence of abnormal B6 levels with 95% confidence and 8% precision. Based on the similarity of study procedures, data availability, and prior analyses, 12 we anticipated obtaining matched control data from 144 participants of the NHANES 2003–2004. 13 A mixed‐effects logistic regression model with group as the fixed effect and pair as the random effect was used to compare the prevalence of B6 abnormalities between the Parkinson's disease and control groups. A linear mixed effects model with group as the fixed effect and pair as the random effect was used to compare the mean log‐transformed B6 level (nmol/L) between the Parkinson's disease and control groups. Because the distribution of B6 levels was right‐skewed, the levels were transformed using the natural logarithm so that they were approximately normally distributed. Estimated means and confidence intervals on the log scale from the linear mixed effects model were exponentiated (back‐transformed) to obtain estimated geometric means on the original scale. Exploratory analyses were performed using linear mixed effects models to examine the associations between demographic variables and log‐transformed B6 level in the Parkinson's disease and control groups; the models included group, the demographic variable of interest, and their interaction as fixed effects, and pair as a random effect. These models yielded estimates of group‐specific associations and estimates of differences in these associations between the groups. Similarly, exploratory analyses were performed using regression models to examine the associations between clinical variables and log‐transformed B6 levels in the Parkinson's disease group. Estimated regression coefficients (β′) for categorical variables represent group differences in means on the log scale. To express these differences on the original B6 scale, the exponentiated coefficients (exp(β′)) are interpreted as the ratio of the geometric mean B6 levels in the two groups. Multiple regression models for associations between demographic variables and log‐transformed B6 level were developed separately in the Parkinson's disease and control groups using backward elimination (α = 0.15) for variable selection. A similar strategy was used to develop a multiple regression model for the association between log‐transformed B6 level and clinical variables in the Parkinson's disease group. These analyses are considered exploratory and not inferential given the variable selection process. Two‐sample t‐tests were used to compare mean log‐transformed B6 levels between people with Parkinson's disease who were and were not using B6 supplementation, and to compare mean log‐transformed homocysteine levels between people with Parkinson's disease with and without polyneuropathy. Pearson correlation coefficients were used to examine associations between log‐transformed homocysteine levels and mTCNS total scores in people with Parkinson's disease, and to examine associations between log‐transformed B6 levels and log‐transformed B12 levels in people with Parkinson's disease and controls. Due to the very small amount of missing data encountered, analyses involving a covariate with missing data were performed using those with complete cases. Results Characteristics of people with Parkinson's disease and controls, including missing data, are summarized in Table 1. Between July 2024 and June 2025, there were 1524 people scheduled for a morning appointment at the URMC, 698 of whom did not have Parkinson's disease or were not scheduled for a follow‐up visit. Sixty‐two people with dementia, three people with known B6 deficiency, and three people with bowel disease (ulcerative colitis, celiac disease, and bowel resection) were excluded. We prospectively and randomly contacted 348 of the remaining 758 potentially eligible people to enroll 144 participants (204 people contacted did not answer or declined to participate). All participants reported overnight fasting at the time of the study visit. Study laboratories reported results of vitamin B12, folic acid, and homocysteine levels within 1–2 days of collection, and results of B6 and methylmalonic acid levels within 1–2 weeks of collection. TABLE 1Characteristics of people with Parkinson's disease and matched controlsParkinson's disease (n = 144)Matched controls a (n = 144)Age, years71 (43–87)71 (43–85)SexFemale55 (38.2%)55 (38.2%)Male89 (61.8%)89 (61.8%)Race White b = 139 (96.5%) Black or African American b = 2 (1.4%) Asian b = 2 (1.4%) Prefer not to answer = 1 (0.7%) Other/Multiracial c = 84 (58.3%) Mexican American c = 28 (19.5%) Non‐Hispanic White c = 21 (14.6%) Missing = 11 (7.6%) Ethnicity Not Hispanic or Latino a = 142 (98.6%) Hispanic or Latino a = 2 (1.4%) Non‐Hispanic Black = 105 (72.9%) Non‐Hispanic White = 28 (19.5%) Mexican American = 11 (7.6%) Body mass index, kg/m2 25.3 (15.9–41.1)27.4 (18.3–48.7)History of diabetesYes20 (13.9%)31 (21.5%)No124 (86.1%)113 (78.5%)History of smokingYes51 (35.4%)76 (52.8%)No93 (64.6%)68 (47.2%)History of alcohol useYes90 (62.5%)92 (63.9%)No54 (37.5%)50 (34.7%)(Missing data)02 (1.4%)History of oral contraceptive use inFemales (n = 55)36 (65.5%)38 (69.1%)Yes19 (34.5%)16 (29.1%)No(Missing data)01 (1.8%)Daily consumption of protein in the last 30 daysYes131 (90.9%)131 (90.9%)No13 (9.1%)13 (9.1%)Daily consumption of food rich in vitamin B6 in last 30 daysYes100 (69.4%)100 (69.4%)No44 (30.6%)44 (30.6%)Food insecurityYes9 (6.2%)22 (15.2%)No135 (93.8%)115 (79.9%)(Missing data)07 (4.9%)Use of any vitamin supplementation in the last 30 daysYes108 (75%)108 (75%)No36 (25%)36 (25%)Use of vitamin B6 supplementation in the last 30 daysData not availableYes11 (7.6%)No133 (92.4%)Age at onset of Parkinson's disease, years64 (36–82)Not applicableDisease duration, years6.7 (1–33)Not applicableLevodopa dosage, mg/day800 (0–2740)Not applicableDuration of treatment with levodopa, months78 (0–312)Not applicable(Missing data)4 (4 participants not taking levodopa)Dosage of dopa decarboxylase inhibitor (carbidopa), mg/day200 (0–685)Not applicableCurrent use of cathecol–o–methyl–transferase inhibitorNot applicableYes14 (9.7%)No130 (90.3%)Levodopa equivalent daily dosage, mg/day800 (0–2720)Not applicableModified Toronto clinical neuropathy score3 (0–22)Not applicablePresence of polyneuropathy (Modified Toronto clinical neuropathy score ≥3)Not applicableYes78 (54.2%)No66 (45.8%)Paroxysmal episodes consistent with epileptic seizuresNot applicableYes3 (2.1%)No137 (95.1%)Uncertain4 (2.8%)Serum level of vitamin B6 (pyridoxal 5′–phosphate), nmol/L27.4 (5–330.5)70.9 (7.1–360.9)Serum level of vitamin B12, pg/mL555 (222–2000)457 (42–3961)Serum level of folic acid, ng/mL14 (4–14)15.8 (3.1–103)(Missing data)2 (2 hemolyzed)0Serum level of homocysteine, μmol/L15 (8–84)10.6 (5.9–53)(Missing data)2 (1 not processed, 1 hemolyzed)0Serum level of methylmalonic acid, μmol/L0.15 (0.1–1.9)0.16 (0.04–2.9)(Missing data)4 (3 not processed, 1 coagulated)10 (reasons unknown) Note: Data are n (%) or median (range). a Matched control data were obtained from the U.S. National Health and Nutrition Examination Survey 2003–2004. b Racial and ethnic categories as defined by the U.S. Office of Management and Budget. c Racial and ethnic categories as reported by the U.S. National Health and Nutrition Examination Survey 2003–2004. Characteristics of people with Parkinson's disease and matched controls White b = 139 (96.5%) Black or African American b = 2 (1.4%) Asian b = 2 (1.4%) Prefer not to answer = 1 (0.7%) Other/Multiracial c = 84 (58.3%) Mexican American c = 28 (19.5%) Non‐Hispanic White c = 21 (14.6%) Missing = 11 (7.6%) Not Hispanic or Latino a = 142 (98.6%) Hispanic or Latino a = 2 (1.4%) Non‐Hispanic Black = 105 (72.9%) Non‐Hispanic White = 28 (19.5%) Mexican American = 11 (7.6%) Note: Data are n (%) or median (range). Matched control data were obtained from the U.S. National Health and Nutrition Examination Survey 2003–2004. Racial and ethnic categories as defined by the U.S. Office of Management and Budget. Racial and ethnic categories as reported by the U.S. National Health and Nutrition Examination Survey 2003–2004. Between July 2025 and December 2025, data were retrieved from the 9179 participants in the NHANES 2003–2004 and all potential matched controls based on age and sex were identified. Controls with documented comorbidities or medication exposures associated with abnormal B6 levels, and those who were not fasting at the time of venipuncture were excluded. The remaining controls were matched to people with Parkinson's disease for consumption of protein, consumption of food rich in B6, and use of vitamin supplementation, with random selection performed if more than one control could be matched with a participant. No data were available from controls on B6 supplementation. Due to lack of control data, we used data from an 85‐year‐old control as a match for an 87‐year‐old participant. Prevalences The estimated prevalence of abnormal B6 levels in people with Parkinson's disease was 49.3% compared to 22.2% in controls; the odds of having abnormal B6 levels were 3.4 times higher than in controls. The estimated prevalence of low B6 levels in people with Parkinson's disease was 36.1% compared to 4.9% in controls; the odds of having low B6 levels were 11.4 times higher than in controls (95% CI = 4.9–26.5, p < 0.001) (Table 2). The geometric mean B6 levels were exp(3.50) = 33.1 nmol/L in the Parkinson's disease group and exp(4.22) = 68.0 nmol/L in controls. The geometric mean B6 level was approximately 49% lower (95% CI = 41–58% lower, p < 0.001) in the Parkinson's disease group than in controls (Fig. 1). TABLE 2Results of mixed effects logistic regression models of the prevalence of vitamin B6 abnormalities in people with Parkinson's disease and matched controlsParkinson's disease (n = 144)Matched controls (n = 144)Odds ratio (95% confidence interval) p‐valueAbnormal serum level of vitamin B6 a No (20–125 nmol/L)73 (50.7%)112 (77.8%)Yes (< 20 or > 125 nmol/L)71 (49.3%)32 (22.2%)3.40 (2.06–5.83)< 0.001Serum level of vitamin B6 b Normal (20–125 nmol/L)73 (50.7%)112 (77.8%)Low (< 20 nmol/L)52 (36.1%)7 (4.9%)11.40 (4.91–26.46)< 0.001High (> 125 nmol/L)19 (13.2%)25 (17.4%)1.17 (0.60–2.27)0.65 Note: Data are n (%). a The analysis used a mixed effects logistic regression model with group as the fixed effect and pair as the random effect. b The analysis used a multinomial mixed effects logistic regression model with group as the fixed effect and pair as the random effect. Results of mixed effects logistic regression models of the prevalence of vitamin B6 abnormalities in people with Parkinson's disease and matched controls Note: Data are n (%). The analysis used a mixed effects logistic regression model with group as the fixed effect and pair as the random effect. The analysis used a multinomial mixed effects logistic regression model with group as the fixed effect and pair as the random effect. Comparison of the distributions of log‐transformed serum level of vitamin B6 in people with Parkinson's disease (red dots) and matched controls (green dots). Log(B6) = Log(serum level of vitamin B6 [nmol/L]). Demographic Associations Results of mixed effects models of associations between B6 levels and demographic variables are presented in Table 3. Results of multiple regression models of associations between B6 levels and demographic variables are presented in Supplemental Table 1. There were associations detected between B6 level and alcohol use and vitamin supplementation in people with Parkinson's disease, and between B6 level and sex in controls. Protein consumption was not analyzed because 131/144 (90.9%) of people with Parkinson's disease reported daily consumption of protein. TABLE 3Results of mixed effects models of associations between serum levels of vitamin B6 (log‐transformed) and demographic variables in people with Parkinson's disease and matched controlsVariableCoefficient (β)95% confidence interval p valueAgeMatched controls−0.007−0.025 to 0.011Parkinson's disease−0.003−0.021 to 0.016Group difference0.0040.68SexSex difference in matched controls (Male–Female)0.5070.219–0.795< 0.001Sex difference in Parkinson's disease (Male–Female)0.175−0.113 to 0.4630.23Group difference−0.3330.06Body mass indexMatched controls−0.001−0.028 to 0.027Parkinson's disease0.018−0.011 to 0.047Group difference0.0190.36History of diabetesMatched controls (Yes–No)−0.148−0.481 to 0.1850.38Parkinson's disease (Yes–No)−0.111−0.507 to 0.2840.58Group difference0.0370.89History of smokingMatched controls (Yes–No)0.222−0.051 to 0.4960.11Parkinson's disease (Yes–No)−0.173−0.459 to 0.1120.23Group difference−0.3960.04History of alcohol useMatched controls (Yes–No)0.108−0.179 to 0.3940.46Parkinson's disease (Yes–No)0.3420.061–0.6230.02Group difference0.2340.24Daily consumption of food rich in vitamin B6 in the last 30 daysMatched controls (Yes–No)0.046−0.264 to 0.3560.77Parkinson's disease (Yes–No)0.181−0.129 to 0.4910.25Group difference0.1350.47Food insecurityMatched controls (Yes–No)−0.209−0.591 to 0.1730.28Parkinson's disease (Yes–No)−0.271−0.838 to 0.2970.35Group difference−0.0620.86Use of vitamin supplementation in the last 30 daysMatched controls (Yes–No)0.129−0.188 to 0.4460.42Parkinson's disease (Yes–No)0.7950.478–1.112< 0.001Group difference0.666< 0.001History of oral contraceptive pill use (in females)Matched controls (Yes–No)0.239−0.269 to 0.7470.35Parkinson's disease (Yes–No)−0.093−0.577 to 0.3900.70Group difference−0.3330.34 Note: The analyses used a mixed effects regression model with fixed effects for group (Parkinson's disease, matched controls), the demographic variable of interest, and the interaction between group and the demographic variable. A random effect of pair was also included. Group difference = Difference in the association parameter (β) between the Parkinson's disease and matched control groups, where β is the slope (for continuous variables) or the difference in means (for yes/no variables). Results of mixed effects models of associations between serum levels of vitamin B6 (log‐transformed) and demographic variables in people with Parkinson's disease and matched controls Note: The analyses used a mixed effects regression model with fixed effects for group (Parkinson's disease, matched controls), the demographic variable of interest, and the interaction between group and the demographic variable. A random effect of pair was also included. Group difference = Difference in the association parameter (β) between the Parkinson's disease and matched control groups, where β is the slope (for continuous variables) or the difference in means (for yes/no variables). Control males had a higher mean log‐transformed B6 level than control females: difference = 0.51 (p < 0.001), corresponding to a 67% increase in geometric mean B6 level (exp(0.51) = 1.67). However, a male–female difference was not detected in the Parkinson's disease group: difference = 0.18 (p = 0.23), corresponding to a 20% increase in geometric mean B6 level (exp(0.18) = 1.20). The group comparison of male–female differences suggests that males had a 39% greater Parkinson's disease‐related reduction in B6 levels (β = −0.33, exp(0.33) = 1.39, p = 0.06). People with Parkinson's disease with a history of alcohol use had a higher mean log‐transformed B6 level than those without history of alcohol use: difference = 0.342 (p = 0.02), corresponding to a 41% increase in geometric mean B6 level (exp(0.342) = 1.41). Vitamin supplement users had a higher mean log‐transformed B6 level than non‐users in the Parkinson's disease group: difference = 0.795 (p < 0.001), corresponding to a 121% increase in geometric mean B6 level (exp(0.795) = 2.21). A difference between vitamin supplement users and non‐users was not apparent in the control group: difference = 0.129 (p = 0.42), corresponding to a 14% increase in geometric mean B6 level (exp(0.129) = 1.14). A comparison of the Parkinson's disease and control groups suggests that vitamin supplementation might mitigate the reduction in B6 levels associated with Parkinson's disease, with supplementation leading to a 95% greater increase in geometric mean B6 levels in the Parkinson's disease group than in controls (group difference = 0.666, exp(0.666) = 1.95, p < 0.001). Clinical Associations Results of regression models of associations between B6 levels and clinical variables in people with Parkinson's disease are presented in Figure 2, Supplemental Table 2 and 3, Supplemental Figure 1 and 2. No associations were detected between B6 level and disease duration or stage, levodopa dosage, treatment duration, or polyneuropathy. There were associations identified between B6 level, MDS‐UPDRS‐IV score, and epileptic seizures. Exploratory associations between log‐transformed serum level of vitamin B6 (“log(B6)” in the Y‐axes) and (A) the MDS‐UPDRS Part IV (X‐axis, range 0–15 points), (B) the presence of epileptic seizures (X‐axis, “No,” “Uncertain” and “Yes” categories), (C) the modified Toronto Clinical Neuropathy Score (mTCNS) (X‐axis, range 0–22 points), and (D) the presence of polyneuropathy (X‐axis, “No” and “Yes” categories) in people with Parkinson's disease. Higher MDS‐UPDRS‐IV scores were associated with lower log‐transformed B6 levels (β = −0.047, p = 0.04, Fig. 2A); a 1–point increase in the MDS‐UPDRS‐IV score was associated with an approximately 5% decrease in geometric mean B6 level (exp(−0.047) = 0.954). Those who had epileptic seizures had a lower mean log‐transformed B6 level than those with no epileptic seizures: difference = −1.650 (p = 0.006, Fig. 2B), corresponding to an 81% decrease in geometric mean B6 level (exp(−1.650) = 0.19). Three of 144 people with Parkinson's disease (2.1%) reported paroxysmal neurological events consistent with epileptic seizures. None of them had a previous diagnosis of epilepsy or seizures. None of them were taking antiepileptic medications. One of them developed medication‐refractory status epilepticus after collecting the blood sample and died before the B6 level was reported (10.6 nmol/L). There was no association detected between B6 level and polyneuropathy (Fig. 2C,D), which was present in 54.2% of people with Parkinson's disease. Additional Analyses People with Parkinson's disease who reported taking B6 supplementation had a higher mean log‐transformed B6 level than those who reported not taking B6 supplementation: difference = 1.21 (p < 0.001), corresponding to a 235% increase in geometric mean B6 level (exp(1.21) = 3.35). People with Parkinson's disease and polyneuropathy had a higher mean log‐transformed homocysteine level than those without polyneuropathy: difference = 0.16 (p < 0.001), corresponding to a 17% increase in geometric mean homocysteine level (exp(0.16) = 1.17). The mTCNS total score was also weakly associated with log‐transformed homocysteine level (r = 0.23, p = 0.007) (Supplemental Fig. 3). No association was detected between B6 and B12 levels in controls (r = 0.08, p = 0.34), but a weak positive association was found between vitamin B6 and B12 levels in people with Parkinson's disease (r = 0.25, p = 0.003) (Supplemental Fig. 4). Discussion In this cross‐sectional study, we found that 52/144 people with Parkinson's disease had low serum levels of vitamin B6 (36.1% prevalence), resulting in >11–fold higher odds of having B6 deficiency when compared to matched controls. These findings confirm that B6 deficiency is highly prevalent in people with Parkinson's disease. Pooled descriptive analyses had previously suggested a 35.9% relative frequency of low B6 levels in people with Parkinson's disease. 5 In a recent cross‐sectional study of 50 participants, 20 the relative frequency of low B6 levels was 14% (7/50). In that study, B6 levels were measured in whole blood, and the fasting status at the time of sampling was not established. Whole blood sampling may yield higher levels than serum sampling because B6, particularly pyridoxal 5′‐phosphate, can accumulate in erythrocytes. 21 , 22 Whole blood and serum levels of B6 can rise transiently after dietary intake. 21 , 22 These considerations highlight the importance of fasting status and specimen type when measuring and interpreting B6 levels. A population‐based study from the NHANES found that B6 levels decrease with age after adolescence, and that the prevalence of low B6 levels is >3% in elderly people. 12 We did not find associations between B6 levels and age, probably because participants were older than the general population. Consistent with the distribution of Parkinson's disease, the male/female ratio in our study was 1.6. In controls, we confirmed that males have higher mean (log‐transformed) B6 levels than females. 12 This sex‐related difference was not detected in people with Parkinson's disease, and the data suggest that males with Parkinson's disease may have greater disease‐related reduction of B6 levels, though this finding is inconclusive. In the NHANES study, low B6 levels were more prevalent in smokers and users of oral contraceptive pills. 12 We found that alcohol use may be associated with higher B6 levels in people with Parkinson's disease. Since we matched people with Parkinson's disease and controls for dietary habits and supplement use, this association may be related to unknown biological or socio‐economic factors. There are complex interactions between B6 and levodopa. Historical observations, reported prior to the introduction of dopa decarboxylase inhibitors, suggested that B6 consumption could reduce levodopa efficacy. 23 , 24 Previously available data suggested an association between relatively high levodopa dosages and low B6 levels. 5 Unexpectedly, we did not find associations between B6 levels and disease duration or stage, dopaminergic medication dosage or levodopa treatment duration. Interestingly, we found an association between higher MDS‐UPDRS‐IV scores and lower B6 levels, suggesting that people with Parkinson's disease and motor complications are at higher risk of B6 deficiency. Future studies should include objective measures of motor complications, assess different levodopa formulations, their cumulative dosages, and examination of interactions between complications, dosage and treatment duration, though these will require larger samples. 3 Three of 144 people with Parkinson's disease had previously undiagnosed epileptic seizures and low B6 levels, and one of them died due to medication‐refractory status epilepticus. Moreover, we found an association between the presence of epileptic seizures and lower B6 levels. Previously available data suggested an association between low B6 levels and epilepsy, 5 including four cases of medication‐refractory status epilepticus. 7 , 8 , 9 , 10 Thirteen additional cases of seizures linked to B6 deficiency in people with Parkinson's disease have been reported through the FDA Adverse Event Reporting System. Thus, the FDA recently issued a drug safety communication regarding seizures associated with low B6 levels in people with Parkinson's disease treated with carbidopa/levodopa. 11 People with Parkinson's disease and motor complications should be monitored for paroxysmal events consistent with epileptic seizures. If seizures are suspected, B6 levels should be obtained as soon as possible. Since laboratory results may be delayed, empiric B6 supplementation should be started immediately after obtaining the blood sample. An antiseizure medication should be started if clinically indicated. Patients with confirmed B6 deficiency may not need long‐term antiseizure medication if seizures resolve and B6 levels normalize with supplementation. Consistent with prior studies, 2 , 6 we found that 78/144 people with Parkinson's disease (54.2%) had polyneuropathy. The presence and severity of polyneuropathy were associated with higher homocysteine levels but not with lower B6 levels. These findings suggest that low B6 levels might not cause polyneuropathy directly but, along with levodopa and other factors, might contribute to homocysteine‐related polyneuropathy in people with Parkinson's disease. 20 , 25 A prior cross‐sectional study found correlations between lower B6 levels and nerve conduction abnormalities, 20 suggesting that electrophysiological markers could detect B6‐related polyneuropathy before symptom onset. 26 Prior studies in people with Parkinson's disease and polyneuropathy suggested complex relationships between B6, B12, and levodopa, potentially including negative associations between B6 and B12 levels in people with Parkinson's disease. 27 , 28 , 29 , 30 However, we found a positive association between B6 and B12 levels in people with Parkinson's disease but not in controls. In another cross‐sectional study, 40/99 people with Parkinson's disease had polyneuropathy. None had low B6 levels and six had high B6 levels despite not taking supplements. 28 , 30 In our study, 108/144 people with Parkinson's disease (75%) reported vitamin supplement use and 11/144 (7.6%) reported B6 supplement use. Since most vitamin supplements contain B6, the latter group was probably taking a B6 dosage higher than recommended. In the NHANES study, 94% of people older than 65 years who consumed more than the recommended daily allowance of B6 also used vitamin supplements. 12 Our exploratory analyses suggest that people with Parkinson's disease and motor complications may benefit from B6 supplementation. However, given the association of B6 toxicity with polyneuropathy, 4 future studies should assess the relationship between B6 supplementation dosage and duration, the development of abnormally high B6 levels and polyneuropathy. This study is limited by recall bias, which we attempted to reduce by sharing structured interviews ahead of time. In future studies, food and supplement intake assessments could be improved by using nutrition research techniques such as food records. Another limitation is that the matched control data were obtained more than 20 years ago. Despite cultural and socio‐economic differences that may impact dietary practices, study procedures were similar and allowed comparisons with the NHANES study focused on B6 levels. 12 Other limitations may arise from the multiplicity of analyses. Thus, the interpretation of the results of the exploratory association analyses should be interpreted with caution. Moreover, 139/144 participants (96.5%) identified as White, potentially limiting the generalizability of the results of this study. In conclusion, vitamin B6 deficiency is highly prevalent in people with Parkinson's disease. Lower B6 levels appear to be associated with motor complications and epileptic seizures possibly resulting in medication‐refractory status epilepticus. Future studies are needed to determine the safety and efficacy of vitamin B6 supplementation to ameliorate motor complications and prevent epilepsy and polyneuropathy in people with Parkinson's disease. Author Roles (1) Research project: A. Conception, B. Organization, C. Execution; (2) Statistical analysis: A. Design, B. Execution, C. Review and critique; (3) Manuscript: A. Writing of the first draft, B. Review and critique. K.J.L.: 1A, 1B, 1C, 2A, 2C, 3A. L.W.: 2A, 2B, 3B. H.M.: 1B, 1C, 3B. N.S.: 1C, 3B. W.B.: 1C, 3B. E.L.: 1A, 2C, 3B. M.P.M.: 1A, 2A, 2C, 3B. A.E.L.: 1A, 2A, 2C, 3B. Disclosures Ethical Compliance Statement: The study was approved by the Research Subjects Review Board of the University of Rochester Medical Center (URMC) on June 6th, 2024 (STUDY00009245). Written informed consent was obtained from all patients participating in the study. We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this work is consistent with those guidelines. Financial Disclosures and Conflicts of Interest: K.J.L. has received compensation for consultancy as course instructor from FHC, and research support through his institution from BlueRock. M.P.M. has received research support through his institution from National Institutes of Health, Food and Drug Administration, Cure SMA, Centers for Disease Control and Prevention, and Dyne Therapeutics, Inc., compensation for consultancy from Abcuro, Inc., compensation for participation on data safety monitoring board or advisory board from National Institutes of Health, Eli Lilly & Company, Neurocrine Biosciences, Inc., ReveraGen BioPharma, Inc., NS Pharma, Inc., Prilenia Therapeutics Development, Ltd., Seelos Therapeutics, Inc., Amgen, Inc., Entrada Therapeutics, Inc., and Appello Pharmaceuticals. A.E.L. has received research support through his institution from Brain Canada, Canadian Institutes of Health Research, Edmond J. Safra Philanthropic Foundation, Krembil Brain Institute, Michael J. Fox Foundation, Parkinson Foundation, Parkinson Canada, and Weston Foundation, compensation for royalties or licenses from Elsevier, Saunders, Wiley‐Blackwell, Johns Hopkins Press, and Cambridge University Press, compensation for consultancy from AbbVie, Amylyx, Aprinoia, Biogen, BioAdvance, Biohaven, BioVie, BlueRock, Denali, EG427, Ferrer, Janssen, Pharma 2B, Sun Pharma, and Ventyx Bio, compensation as speaker honoraria from Sun Pharma, and AbbVie, and compensation as payment for expert testimony for litigation related to paraquat and Parkinson's disease. L.W., H.M., N.S., W.B., and E.L. report no disclosures. Author disclosures are available in the Supporting Information. Supporting information Supplemental TABLE 1 Results of multiple regression models of associations between serum levels of vitamin B6 (log‐transformed) and demographic variables in people with Parkinson's disease and matched controls Supplemental TABLE 2 Results of regression models of associations between serum levels of vitamin B6 (log‐transformed) and clinical variables in people with Parkinson's disease Supplemental TABLE 3 Results of multiple regression models of associations between serum levels of vitamin B6 (log‐transformed) and clinical variables in people with Parkinson's disease Supplemental Figure 1 Results of regression models of associations between serum levels of vitamin B6 (log‐transformed) and categorical clinical variables in people with Parkinson's disease. Supplemental Figure 2 Results of regression models of associations between serum levels of vitamin B6 (log‐transformed) and quantitative clinical variables in people with Parkinson's disease. Supplemental Figure 3 Relationships between serum levels of homocysteine (log‐transformed) and polyneuropathy and the modified Toronto clinical neuropathy score. Supplemental Figure 4 Relationship between serum levels of vitamin B6 (log‐transformed) and serum levels of vitamin B12 (log‐transformed) in people with Parkinson's disease and matched controls. Data S1. Disclosure Statement form.