Covariation between plasma phosphate and daytime cortisol in early Parkinson's disease

Abstract Background Disturbed phosphate homeostasis in early Parkinson′s disease (PD) may originate from a stress‐related condition and nutritional status among other risk factors, age, and gender. Methods Risk of malnutrition using Mini‐nutritional assessment (MNA score) and plasma levels of protein markers and daytime cortisol at the time of diagnosis in PD (n = 75) were compared with a control group (n = 24). Cognition was assessed using the Mini‐Mental State Examination (MMSE score) and motor function using Unified Parkinson′s Disease Rating Scale (UPDRS‐part III scale). Results The patients with PD had significantly lower MNA score than controls which correlated with plasma phosphate levels. The logistic regression revealed that increasing MNA protected from low plasma phosphate, final score (OR = 0.399; 95% CI = 0.196–0.816; p = .012) and total score (OR = 0.656; 95% CI = 0.422–1.018; p = .060). Phosphate correlated with albumin (r = .315; p < .006), transferrin (r = .331; p < .004) and cortisol (r = −0.355; p < .002) confirmed by logistic regressions. Increasing albumin protects from low phosphate after adjusting in logistic regression (OR = 0.806; 95% CI = 0.682–0.952; p = .011) and after including variables from Table 1 in backwards elimination, final step (OR = 0.800; 95% CI = 0.660–0.969; p = .022). MNA total score and cortisol correlated inversely, confirmed in logistic regression for MNA total score (OR = 0.786; 95% CI = 0.627–0.985; p = .037) and for MNA initial score (OR = 0.650; 95% CI = 0.453–0.930; p = .020). Conclusion This study highlights the importance of phosphate for optimal nutritional status by association with MNA score and albumin in plasma. An inverse relationship between phosphate and cortisol indicate, in addition, that low phosphate levels may affect cognition and motor function in PD.


Covariation between plasma phosphate and daytime cortisol in early Parkinson's disease Lena Håglin | Lennart Bäckman 1 | INTRODUCTION
Studying links between cognition, motor disturbance, and malnutrition in patients with Parkinson′s disease (PD) require understanding of how biomarkers for protein balance and metabolism may be affected by stress and diet. Protein energy malnutrition (PEM) may increase risk of multiple nutrient deficiencies, which in addition to body weight loss and medication, may contribute to both cognition deficits and motor disturbances in PD.
Patients with PD have higher blood cortisol levels than controls, associated with an increased acrophase, increased amplitude, and increased area under the curve of cortisol (Charlett et al., 1998;Hartman, Vedldhuis, Deuschle, Standhardt, & Heuser, 1997). In PD, nonmotor as well as motor symptoms can result from dysregulation within the HPAaxis, including high levels of cortisol that can have deleterious effects on multiple organ systems. Acute levodopa medication can result in low HPA-axis activity with a decrease in cortisol levels, most pronounced in patients with severe disabilities (Müller, Welnic, & Muhlack, 2007).
As cortisol is a phosphaturic hormone, loss of phosphate may be the cause of low plasma levels and/or body depletion of high-energy phosphate in intracellular compartments associated with PD.
Phosphate depletion may cause anorexia, weight loss, and disturbed energy metabolism with hypoxia in neurological and neuropsychiatric diseases and symptoms (Håglin, 2016). A schematic model for stressinduced PD suggests that glucocorticoids can be neurotoxic (Smith, Castro, & Zigmond, 2002). Glucocorticoids and chronic stress influence the progress of PD by putting neurons in an energetically unfavorable condition (Kibel & Drenjančević-Perić, 2008). Salivary cortisol concentration is decreased in patients with PD after tactile massage, a finding that suggests a connection between stress and cortisol levels in this patient population (Törnhage et al., 2013). Salivary cortisol correlates with risk behavior in PD (Djamshidian et al., 2011).
In this study, we examine associations between nutritional status, phosphate, albumin, and transferrin together with cortisol levels,   (Gibb & Lees, 1988). Patients fulfilling diagnostic criteria for PD were included in the NYPUM study. A control group (n = 30) was selected by announcements, according to the age and gender of the first 50 patients diagnosed in the NYPUM study.

| NYPUM population
The present substudy included patients with PD (n = 75) and controls (n = 24) who took part in the nutritional status assessment at baseline of the NYPUM project. In addition, the baseline data included analysis of nutrition-related biomarkers in a blood sample (Table 1).

| Disease severity
Disease severity (i.e., the motor symptom severity) was assessed using the UPDRS total score and UPDRS-part III scale (Fahn, Elton, &, The UPDRS Development Committee, 1987) and the Hoehn and Yahr staging scale (Hoehn & Yahr, 1967). The UPDRS scoring was performed when the patients were in the ON-phase and when they started the dopaminergic treatment. The results from the UPDRSpart III scale were used to study associations with nutritional status, biomarkers, and cognition. Cognitive function was evaluated using the MMSE score, a screening instrument for global cognition, including orientation in time and place, ability to follow simple commands, registration, attention, calculation, memory, naming, writing, and figure copying (Folstein, Folstein, & Hugh, 1975). MMSE score ranges from 0 to 30, with higher scores indicating better cognitive functioning.

| Anti-Parkinson medication
The levodopa equivalent dose (LED) was calculated at baseline and at follow ups, by use of a conversion factor for each of the anti-Parkinson T A B L E 1 Patient baseline characteristics and plasma variables in mean (min-max) and with p-values for difference between PD patients and controls medication (Tomlinson et al., 2010). Information about LED at baseline revealed that only two patients had started anti-Parkinson treatment at the time of diagnosis. At first follow up at 6 month, all but nine patients had started treatment and about 50% of the patients had a LED <200 and 8% had >300.
The medication was introduced over the first weeks after diagnosis. The nutritional investigation, including the blood sampling, was performed in a nonstandardized manner according to drug-therapy during this time period. We found that 13 of the patients started anti-Parkinson medication prior to the blood sampling, whereas 60 patients started afterward. The mean levels of plasma cortisol in these groups were 412 ± 165 nmol/L (n = 13) and 385 ± 139 nmol/L (n = 60), respectively; p < .537.

| Mini-Nutritional Assessment
The Mini-Nutritional Assessment (MNA) score, an international validated screening tool, was used to assess nutritional status and was performed when anthropometrical measurements were gathered by the dietician. The screening consists of 18 questions regarding anthropometry, diet, and health (Guigoz, 2006). MNA-total scores between 24 and 30 points indicate optimal nutritional status, MNA total scores between 17 and 23.5 indicate a risk for malnutrition, and MNA scores <17 indicate malnutrition. Initial questions refer to MNA-initial (MNA-SF; short form) in this study with score from 7 to 14 points.
The last items of the MNA score included questions about food, selfperceived health, and anthropometry (MNA final score). The scores for this information were between 10 and 16 points.

| Biochemical analysis
The blood was collected between 8 a.m. and 4 p.m. on the day when the patients were admitted for assessment of nutritional status, dietary intake, and neuro-psychological tests, corresponding to 2-6 weeks after diagnosis. The collection of nutritional data including the blood sampling was performed before initiation of anti-Parkinson medication in the majority of the patients (N = 60 patients).

| Data analysis
An independent two-tailed test was used to compare patients and controls. For non-normal distributed data, a chi-square test was used.
Bivariate correlations were analyzed using the Pearson's correlation coefficient (Tabachnik & Fidell, 2001 Written informed consent was obtained from all participants.

| Multiple logistic regressions
Adjustments were done for age and gender in the two models with either plasma phosphate (low level = 1; 0.73-1.20 and high level = 0; 1.21-1.53) or plasma cortisol (high level = 368.7-778.5 and low level = 147.6-366.8) as the dependent variable (Table 3).
The results revealed that increasing cortisol is a risk for low phos-

| Backwards elimination
The final step of backwards elimination reveals that increasing cortisol increases risk for low phosphate while high albumin decreases risk of low phosphate (Table 4). Gender and MNA final score was also confirmed in this regression. To be women and to increase MNA score, protects from low phosphate levels (Table 4).
The final step of backwards elimination reveals that increasing albumin or UPDRS-part III scale, increases risk for high cortisol while increasing MNA total score, plasma phosphate or UPDRS-total decreases this risk (Table 4).

| DISCUSSION
The plasma phosphate concentration was positively associated with plasma albumin, transferrin, and MNA score at the time of diagnosis, indicating an involvement in nutritional status (Fig. 1) high cortisol levels-a stress-induced catabolic response frequently reported as weight loss. Increase in cortisol may exacerbate disease progress by loss of weight and skeletal muscle and increased visceral fat and thereby cause sarcopenic obesity. In a study on gait difficulties in PD, the level of cortisol was similar to the level found in the present patient population (Charlett et al., 1998). This finding can be either due to a stress-related disturbance in Parkinsons′s disease, shown by higher levels of cortisol, that causes low phosphate levels or be due to malnutrition with disturbed protein status with high cortisol.
Low albumin and low phosphate have been shown to be associated in geriatric patients (Kagansky, Levy, Koren-Morag, Berger, & Knobler, 2005) and in a healthy population (Wojcicki, 2013). Thus, a low level of phosphate may affect protein metabolism negatively. It has been reported that visceral proteins could be a useful index in detecting malnutrition among elderly (Kagansky et al., 2005;Sergi et al., 2006).
In clinical practice, transferrin may be used as an indirect marker of PEM, but transferrin is influenced by many conditions associated with old age and in this study, negatively associated with plasma cortisol in the bivariate analysis. Transferrin may respond as an acute phase reactant (short term compared with albumin) and thus also be an indication of acute severe condition (stress) rather than chronic malnutrition. In addition, transferrin increases with iron deficiency, a condition present in some cases of malnutrition (Ingenbleek, Van Den Schrieck, De Nayer, & De Visscher, 1975).
Low phosphate levels has been neglected in clinical studies, as an important biomarker for malnutrition, whereas low levels of albumin and transferrin have been used in prognostic indexes of chronic malnutrition and transferrin as a marker for negative nitrogen balance and visceral protein depletion (Ingenbleek et al., 1975). Independent of causes, low phosphate levels have detrimental clinical outcome in diseases and is associated with morbidity and mortality among elderly with a long hospital stay and lower albumin levels (Kagansky et al., 2005;Sumukadas et al., 2009).
Phosphate and its importance for energy generation, anabolism, and acid-base maintenance have been neglected with respect to studies in neurological diseases. Brain bioenergetics has impact on cognitive ability and various Pi containing ratios correlated with cognition in both adults and children/adolescents (Heinze et al., 2015).
Hattingen et al. demonstrated mitochondrial dysfunction in early PD (Hattingen et al., 2009). Using PET scanning and metabolic ratio, Liepelt et al. (2009) found an association between hypometabolism and both UPDRS-part III scale and MMSE score and concluded that cerebral hypometabolism in PD is primarily associated with cognitive impairment.
It has been shown in an experimental study that stress by high cortisol exaggerates motor symptoms in a rat model of PD (Smith, Jadavji, Colwell, Perehudoff, & Metz, 2008). Clinical studies have reported that high levels of cortisol can have negative effects on some cognitive functions that are measured during day time (Comijs et al., 2010). A high level of the MMSE score was associated with high plasma phosphate and high MNA score, indicating good nutritional status and high level of cognition with low cortisol levels during daytime.
However, among elderly, low cortisol levels in the evenings are associated with depression indicating influence from diurnal variation on T A B L E 4 Variables included from Table 1 for PD patients in a multiple logistic regression with either plasma phosphate (low level = 1; 0.73-1.20 and high level = 0; 1.21-1.53) or plasma cortisol (high level = 1; 368.7-778.5 and low level = 0; 147. F I G U R E 1 Low plasma phosphate has consequences for energy metabolism in almost all cells in a body and thereby a limiting factor for optimal cognition and motor function. The data presented in Table 2 indicates that the interactions and possible sequences of these events are correlated and can be expressed as two theories; (A) stress-related decrease in phosphate and transferrin induced by cortisol and (B) protein energy malnutrition with decrease in phosphate and albumin revealed by low MNA score. Plasma biomarkers may explain disease severity assessed with UPDRSpart III scale and MMSE score, respectively, through low plasma phosphate level
On the contrary, the negative relationship between UPDRS-part III scale and MMSE-score in the logistic regression (Table 3) and between UPDRS-part III scale and cortisol, MMSE and phosphate (Table 2) may reveal adverse effects and some mechanism behind the covariance between phosphate and cortisol on motor dysfunction (Fig. 1A). It has been shown that acute L-dopa administration induces a decrease in cortisol by a decreased function of HPA-axis which is more pronounced in advanced stage of the disease (Müller et al., 2007). Patients who started anti-Parkinson medication before the blood sampling for cortisol analysis did not have lower cortisol than patients who received the medication after cortisol assessments. Thus, the anti-Parkinson medication was not a confounder for conclusions about negative effects from cortisol on nutritional status and cognition.
We conclude that cognitive disturbance is associated with low phosphate plasma due to protein malnutrition in patients at the time of diagnosis of PD (Fig. 1B). The results reveal the importance of including both phosphate and cortisol in future studies that examine nutritional status, cognition, and motor function in PD and for establishing new knowledge about mitochondrial energy production in the brain (Heinze et al., 2015). Low levels of plasma phosphate have never before been discussed as a metabolic marker and related to disease symptoms in the PD syndrome. A negative phosphate balance either due to low intake or losses due to stress and high cortisol may have multiple consequences for the disease pathogenesis of early PD.