Osteopontin as an early predictor of atherosclerosis in attack-free Familial Mediterranean fever patients

Abstract

Familial Mediterranean fever (FMF) is an autoinflammatory disease that is associated with endothelial dysfunction and atherosclerosis. Osteopontin which is a multifunctional protein involved in the modulation of inflammatory processes may contribute to the development of atherosclerosis in FMF patients. Therefore, this cross-sectional study investigated the relationship of osteopontin with carotid intima media thickness (CIMT) and atherogenic indices in patients with FMF. Serum osteopontin levels, CIMT, Castelli risk index I and II, plasma atherogenic index (PAI), non - high-density lipoprotein cholesterol, and atherogenic coefficient (AC) in 64 attack-free FMF patients were compared with levels in 23 healthy control subjects. The serum osteopontin level, CIMT, Castelli risk index I, AC and PAI were significantly higher, and high-density lipoprotein cholesterol was significantly lower in FMF patients (P < .001, P < .001, P = .045, P = .016, P = .045, and P = .024; respectively). There were significant positive correlations between osteopontin and CIMT, PAI, AC, and Castelli risk index I (R = 0.580, R = 0.259, R = 0.233, R = 0.277; respectively) and there was significant negative correlation between osteopontin and high-density lipoprotein cholesterol (r= −0.309). Patients who had homozygote mutations had significantly higher osteopontin, PAI, Castelli risk index I and II level. The current study is the first to demonstrate significantly increased serum osteopontin levels in attack-free FMF patients compared with healthy controls. It was also associated with CIMT and many atherogenic indices. This finding provides a new experimental basis to understand the pathogenesis of inflammation-induced atherosclerosis in FMF patients. Furthermore, patients who had homozygote mutations had worse atherogenic indices than those with heterozygote mutations.

1. Introduction

Familial Mediterranean fever (FMF) is a genetic autoinflammatory disease that is characterized by recurrent inflammatory attacks of the serosal and synovial membranes.^[1] Subclinical inflammation may persist even during the absence of symptoms in FMF patients.^[2] The ongoing subclinical inflammation brings about endothelial dysfunction which finally causes atherosclerosis.^[3] The process continues with vascular morphology changes and then with negative cardiovascular outcomes.^[4] However, the relationship between subclinical inflammation and atherosclerosis in attack-free FMF patients has not been clearly demonstrated until now. Various changes in a gene called the mediterranean fever (MEFV) are linked to FMF.^[5] Some variations may cause more severe inflammation and symptoms, while others may cause milder disease.^[6–8] However, there is no information in the literature about the effect of different MEFV genotypes on atherosclerosis.

Carotid intima media thickness (CIMT) measurement and evaluation of the atherogenic indices which is calculated according to conventional lipid panel are some of the methods which are used to evaluate atherosclerosis. CIMT is an ultrasound screening test for evaluation of the presence and extent of atherosclerosis. Atherosclerosis is also associated with atherogenic dyslipidemia, which composes the increased serum triglyceride (TG) and small dense low-density lipoprotein cholesterol (LDLc) levels along with decreased high-density lipoprotein cholesterol (HDLc) level.^[9,10] Recent studies have demonstrated that, the use of atherogenic indices has better predictive capability for atherosclerosis and are less susceptible to disease activity variation during large periods of time as compared with concentrations of conventional atherogenic dyslipidemia panel.^[11,12] Furthermore, analysis of inflammatory mediators associated with cardiovascular disease can be used to determine the atherosclerosis. Osteopontin is a multifunctional phosphoprotein and acts as a pro-inflammatory cytokine. The previous data indicated that increased concentration of osteopontin contributes to inflammatory processes and atherosclerosis.^[13,14] Therefore, plasma osteopontin levels might serve as a sensitive biomarker for atherosclerosis. Many modifiable risk factors for atherosclerosis have also been identified. One of them is vitamin D deficiency which is associated with endothelial dysfunction and atherosclerosis in healthy humans and different patient populations.^[15,16]

Despite its emerging role in cardiovascular disease, the severity and importance of atherosclerosis in attack-free FMF patients is not clear. Therefore, we aimed to investigate the atherosclerosis with measurement of CIMT, atherogenic indices, vitamin D and osteopontin in attack-free FMF patients who had different MEFV genotypes and compare them to those of healthy controls.

2. Materials and Methods

2.1. Patients and clinical assessment

This is a single-center cross-sectional study in patients admitted to the internal medicine outpatient clinic of a tertiary research hospital between February 2020 and August 2021 who were on follow-up with FMF diagnosis. The control group included age- and sex-matched healthy volunteers who had no clinical signs of FMF or any other disease and no family history of FMF. This study was performed in accordance with the Helsinki Declaration and approved by the local ethics committee (Ethics Committee approval code: 2019/426). We obtained written informed consent from all patients.

Demographic data, duration of symptoms, age at diagnosis, frequency and duration of FMF attacks, response to colchicine, and determined MEFV mutation types were recorded. The ratio of weight to the square of height (kg/m2) was given as body mass index.

FMF patients were included if fulfill clinical diagnostic criteria for FMF and between 18 and 50 years regardless of sex, duration of FMF and age of onset. The diagnosis of FMF was established according to the Tel-Hashomer criteria: the presence of at least 1 of 4 major criteria, 2 of 5 minor criteria, 1 minor criterion plus 5 of 10 supportive criteria, or 4 of 5 specific supportive criteria.^[17] Patients included fulfilled 2 major or 1 major and 2 minor criteria. FMF patients had been receiving only maintenance doses of colchicine (1–1.5 mg/day). No patients had acute attacks at the time of investigation.

The exclusion criteria were as follows: patients with diabetes mellitus, metabolic syndrome, hypertension, impaired renal or thyroid function, active infectious disease, intestinal, liver, musculoskeletal or skin diseases, coronary artery disease and malignancy, patients using alcohol and/or smoking, patients who were pregnant, and patients taking any medication other than colchicine. The patients who had last attack within 1-month period were excluded from the study.

2.2. Laboratory analysis

An autoanalyzer was used to obtain serum lipid profile including TG, LDLc, HDLc, and total cholesterol. The atherogenic indices including Castelli risk index I and II (total cholesterol/HDLc and LDLc/HDLc; respectively), plasma atherogenic index (PAI; log TG/HDLc), non – high-density lipoprotein cholesterol (non-HDLc) (TG-HDLc) and atherogenic coefficient (AC; non-HDLc/HDLc) were calculated.^[9,11]

At the end of a minimum 12 hour fasting, blood samples were collected from FMF patients and healthy controls. A complete blood count and biochemical analyses including C-reactive protein (CRP), erythrocyte sedimentation rate, fasting blood glucose creatinine, albumin, aspartate transaminase, alanine transaminase, uric acid, thyroid stimulating hormone, 25 (OH) vitamin D, vitamin B12, folic acid and urine microprotein/creatinine ratio were studied on the day of blood collection.

Serum specimens of patients and controls for analysis of osteopontin were stored at −80°C until assayed. Serum osteopontin level was studied by enzyme linked immunosorbent assay using a commercial kit (Limit of Quantification: 0, 31 ng/mL, Elabscience, USA). Osteopontin level was analyzed according to the manufacturer’s instructions and expressed as ng/mL. The concentrations of the samples were calculated through calibration curves obtained from study standards with known levels. The regression coefficients of our calibration curve was R = 0,997. Intraassay coefficient of variation of the assay is < 10%, interassay coefficient of variation of the assay is < 10% for both parameters. CRP was analyzed with Cobas c702 (Cobas, Roche Diagnostics Mannheim, Germany) with immunoturbidimetric assay. CRP level was expressed as mg/L.

2.3. Genetic analysis

Genomic DNA was extracted with the DNA isolation kit (Zinexts Life Science Corporations, Taiwan) from peripheral blood samples. Next-generation sequencing was conducted with the QIAseq Targeted DNA panel. Sequencing reactions were performed with the MiSeq next-generation sequencing system (Illumina, USA). The MEFV gene (RefSeq transcript NM_000243.3; panel code: CDHS-10949Z-36) was set as the target gene. Bioinformatics analysis was performed using 1 or more of the QIAGEN Clinical Insight Analyze, QIAGEN Clinical Insight Analyze Universal, and QIAGEN Clinical Insight Interpret interfaces (https://digitalinsights.qiagen.com/), and variants that were found significant were listed in the report. Genetic analysis data of FMF patients were obtained from hospital database.

2.4. Ultrasonography measure

The CIMT was measured at supine position with slight cervical extension using high-resolution 7.5 Mhz linear probe (Hitachi EUB 6500, Osaka, Japan). The measurements were performed at a point 10-mm away from internal carotid artery and carotid artery bifurcation using 2-dimensional sonography. To minimize effects of arterial compliance on results, cardiac monitorization and peak-R wave coupling (for correlation with each phase of cardiac cycle) were used. Measurements were made at 3 different sites in each session. Mean CIMT was defined average of 6 measurements obtained in 2 different sessions. The CIMT ≥ 0.9 mm was defined as abnormal.

2.5. Statistical analysis

The normality and the homogeneity of the data were evaluated by the Kolmogorov–Smirnov test and Levene test, respectively. Comparisons between groups for continuous variables were performed using the Student t test for normal distribution or the Mann–Whitney U test for non-normal distribution. Fisher test or the Chi-square test was used for all categorical data. The Pearson or Spearman correlation analysis was performed to determine the relationship between osteopontin and other parameters according to the distribution of parameters. Statistical analyses were performed using SPSS version 22.0 (IBM Corp., Armonk, NY). Data were presented as mean ± standard deviation, the median (interquartile range) or count (percent). P < .05 was considered statistically significant.

3. Results

We included 64 patients with FMF and 23 healthy controls in this study. In the FMF group, 16 were male (25%) and 48 were female (75%), with a mean age of 33.8 ± 8.4 years. The mean duration of disease was 2.5 ± 0.3 years. The most common MEFV mutations in FMF patients were R202Q heterozygote (n = 26), M694V heterozygote (n = 19) and V726A heterozygote (n = 14) mutations. The frequency of MEFV mutations were shown in Table 1. Five patients had both homozygote and heterozygote mutations and 31 patients had multipl heterozygote mutations without homozygote mutations.

Table 1.

Mutation types in FMF patients.

Mutations	Mutation type	n (%)
Homozygote	R202Q	6 (9.4)
	M694V	4 (6.25)
	M680I	3 (4.7)
Heterozygote	R202Q	26 (40.6)
	M694V	19 (29.7)
	V726A	14 (21.9)
	E148Q	9 (14.1)
	M680I	9 (14.1)
	P369S	4 (6.25)
	R408Q	3 (4.7)
	F479I	3 (4.7)
	R761H	1 (1.6)
	E167D	1 (1.6)
	K695R	1 (1.6)
	G304R	1 (1.6)
	Y471X	1 (1.6)
	I591T	1 (1.6)
	C226Y	1 (1.6)

FMF = Familial Mediterranean fever.

The demographic data of the FMF patients and the control group were shown in Table 2. There were no significant differences in sex, age-and body mass index between groups. There were significant differences between FMF patients and healthy controls with respect to serum osteopontin level and HDLc (P < .001 and P = .024; respectively). The serum osteopontin level was significantly higher and HDLc was significantly lower in FMF patients. However, there were no significant differences in other laboratory measurements and lipid profile between groups.

Table 2.

Demographic and laboratory data of FMF patients and control groups.

Variables	FMF patients (n = 64)	Control group (n = 23)	P value
Gender, M/F	16 (25)/ 48 (75)	6 (26.1)/ 17 (73.9)	.918
Age, yr	33.8 ± 8.4	31 ± 7.3	.168
BMI, kg/m2	24.6 (4.3)	24.8 (4.8)	.900
Osteopontin, ng/mL	31.8 (21.2)	18.9 (9.6)	<.001
C-reactive protein, mg/L	1.7 (3)	1.5 (1.5)	.149
White blood cell count, 103/µL	7.4 ± 1.6	7.4 ± 1.5	.921
Hemoglobin, g/dL	13.8 ± 1.8	14.4 ± 1.4	.174
Erythrocyte sedimentation rate, mm/h	7 (9)	6 (8)	.389
AST, U/I	18 (6)	16 (6)	.337
ALT, U/I	18.5 (15)	15 (9)	.897
Platelets, 103/µL	285 (88)	254 (86)	.215
Creatinin, mg/dL	0.68 (0.17)	0.72 (0.15)	.113
Fasting plasma glucose, mg/dL	89 (11)	90 (9)	.287
Lipid profile
LDLc, mg/dL	97 (44)	95 (41)	.555
Triglycerides, mg/dL	110 (67)	90 (53)	.160
HDLc, mg/dL	46 (19)	59 (20)	.024
Total cholesterol, mg/dL	171 (47)	168 (26)	.418
Albumin, g/L	4.63 ± 0.3	4.68 ± 0.4	.547
Uric acid, mg/dL	4.42 ± 1.1	4.46 ± 1.1	.882
Thyroid stimulating hormone, mU/L	1.97 ± 0.9	1.94 ± 0.9	.872
Vitamin B12, µg/d	314 (140)	315 (95)	.716
Folate, µg/L	6.4 (3)	7.7 (4)	.057
Spot urine microprotein/ creatinin ratio, ug/mg creatinin	63.8 (31)	56.9 (33)	.591
Vitamin D, µg/d	12.6 (8)	15.8 (11)	.119

Data expressed as mean ± standard deviation, median (interquartile range) or number (percentage).

ALT = alanine aminotransferase, AST = aspartate aminotransferase, BMI = body mass index, FMF = Familial Mediterranean fever, HDLc = high-density lipoprotein cholesterol, LDLc = low-density lipoprotein cholesterol.

Table 3 presents atherogenic indices in FMF patients and controls. Three of the 5 atherogenic indices were significantly higher in FMF patients (Castelli risk index I, AC and PAI; P = .038, P = .038 and P = .045; respectively). The median CIMT measurements were also significantly higher in FMF patients than controls (P < .001).

Table 3.

Atherogenic indices and CIMT in FMF patients and controls.

	FMF patients (n = 64)	Control group (n = 23)	P value
Castelli risk index I	3.9 (1.2)	2.8 (1.3)	.038
Castelli risk index II	2.2 (1)	1.6 (1.3)	.062
Non-HDLc	122 (49)	110 (32)	.087
AC	2.8 (1.2)	1.8 (1.3)	.038
PAI	0.4 (0.4)	0.2 (0.4)	.045
CIMT, mm	0.64 (0.22)	0.4 (0.08)	<.001

Data expressed as median (interquartile range).

AC = atherogenic coefficient, CIMT = carotid intima media thickness, FMF = Familial Mediterranean fever, HDLc = high-density lipoprotein cholesterol, PAI = plasma atherogenic index.

Correlation analyses revealed that osteopontin had significant relationship with CIMT, PAI, AC, and Castelli risk index I (R = 0.580, P < .001; R = 0.259, P = .018; R = 0.316, P = .004; R = 0.316, P = .004; respectively). Furthermore, there was a significant negative correlation between osteopontin and HDLc (r= −0.309, P = .004). Scatter plot graphs of significant correlation between osteopontin and CIMT and HDLc were showed in Figure 1. However, we found no significant correlation between remaining lipid parameters, CRP and osteopontin. In our study, it was also examined whether there was a correlation between vitamin D and any parameter of atherosclerosis which used in this study. We did not determine a significant correlation between vitamin D and osteopontin, CIMT, HDLc, PAI, AC, Castelli risk index I and II (R = 0.089, P = .547; r = −0.050, P = 735; r = −0.009, P = .952; R = 0.087, P = .556; R = 0.115, P = .435; R = 0.115, P = .435; R = 0.108, P = .467; respectively).

Figure 1.

Scatter plot graphs of correlation of osteopontin with CIMT and HDLc (R = 0.580, P < .001 and r= −0.309, P = .004; respectively). CIMT = carotid intima media thickness, HDLc = high-density lipoprotein cholesterol.

MEFV mutation types of patients were evaluated according to atheroslerosis indices. The highest level of PAI and Castelli risk index I were detected in patients who had R202Q homozygote mutation. The highest level of AC and non-HDLc were found in patients who had M680I homozygote mutation. The highest value of osteopontin and CIMT was detected in patients who had M694V homozygote mutation (Figure S1 and S2, Supplemental Digital Content, http://links.lww.com/MD/J731, http://links.lww.com/MD/J733). These results suggest that those with homozygote mutations have worse atherogenic indices. We confirmed this result by dividing the FMF patients into 2 groups according to their mutations. Patients who had homozygote mutations had significantly higher osteopontin, PAI, AC, Castelli risk index I and II level (Table 4). Although not statistically significant, CIMT also tends to be higher in those with homozygote mutations. There was no significant difference in level of osteopontin, CIMT, PAI, non-HDLc, AC, Castelli risk index I and II between patients with multipl mutations and 1 mutation (Table S1, Supplemental Digital Content, http://links.lww.com/MD/J735).

Table 4.

Comparison of atherogenic indices, osteopontin and CIMT in FMF patients who had homozygote and heterozygote mutations.

	Patients with homozygote mutations (n = 12)	Patients with heterozygote mutations (n = 52)	P value
Osteopontin, ng/mL	47 (46)	25 (18)	<.001
CIMT, mm	0.7 (0.4)	0.6 (0.2)	.366
PAI	0.47 (0.5)	0.35 (0.4)	.026
Non-HDLc	121 (52)	122 (47)	.724
AC	3.1 (0.8)	2.4 (1.3)	.013
Castelli risk index I	4.1 (0.8)	3.4 (1.3)	.013
Castelli risk index II	2.69 (0.5)	2.1 (1)	.020

Data expressed as median (interquartile range).

AC = atherogenic coefficient, CIMT = carotid intima media thickness, FMF = Familial Mediterranean fever, HDLc = high-density lipoprotein cholesterol, PAI = plasma atherogenic index.

4. Discussion

Screening of atherosclerosis can help to identify or predict the risk of cardiovascular disease and the features associated with atherosclerosis including complications and death. In the present study, we measured serum osteopontin levels in addition to atherogenic indices and CIMT in attack-free FMF patients. Thus, we mainly focused on evaluating whether early atherosclerosis is present even in the absence of attack in FMF patients. There was significant difference in the level of some atherogenic indices, CIMT and serum osteopontin level between the attack-free FMF patients and the healthy controls. Although there are studies which showed the relationship between FMF and atherosclerosis measured with some atherogenic indices,^[18,19] we have confirmed this relationship with serum osteopontin levels for the first time in the literature. Moreover, it was very important that we demonstrated this relationship in attack-free patients. In present study, the serum osteopontin level, CIMT and the levels of Castelli risk index I, AC and PAI were significantly higher and HDLc was significantly lower in FMF patients. Furthermore, this study is the first evaluation of the association of MEFV mutation types with atherosclerosis. Those with homozygote mutations had poor features in terms of atherosclerosis.

In the course of FMF, decreased HDLc values is the most frequently observed lipid disturbance in especially patients receiving colchicine treatment.^[18,20,21] Similarly, the levels of HDLc was significantly lower in our FMF patients than healthy controls, while other lipid parameters did not differ between both groups. Many studies demonstrated that atherogenic indices may better determine the risk level for cardiovasculary disease than any other single lipid parameter.^[22,23] The role of atherogenic indices was investigated in FMF patients in many different studies. Castelli risk index I was significantly higher in patients with FMF or ankylosing spondylitis and associated with presence of endothelial dysfunction and atherogenesis which evaluated with flow-mediated dilatation.^[24] Other study showed a positive correlation between PAI and CIMT in patients with FMF.^[25] Acay et al^[18] and Cakirca et al reported similar results that PAI, AC, Castelli risk index I and II were higher in FMF patients.^[26] Our findings were in parallel with these reports. We also strengthened these results with osteopontin measurement. Thus, the relationship between atherosclerosis and FMF was revealed more clearly.

In a study, the most frequent mutations of the MEFV gene were E148Q, M680I, M694V, and V726A mutations in Turkish population.^[27] However in our study, homozygote and heterozygote R202Q mutations were the most common mutations. The potential implications of this quite new result need to be evaluated in future studies. It is known that homozygote mutation in the M694V gene is linked with a more severe clinical course.^[28] However, the relationship between atherosclerosis and MEFV mutations, especially regarding the role of homozygote or heterozygote mutations, have not been studied sufficiently in the literature. One study did not find significant association between M694V mutation and lipid profile and PAI.^[18] Ozel et al^[29] showed the association between MEFV gene homozygote mutation and inflammatory complications. But in other study, there were no significant differences in CIMT, lipoprotein-a, homocysteine and other inflammatory parameters between patients with M694V homozygote mutations and patients with any other heterozygote mutations.^[30] The discrepancy between these studies may be related to the fact that the studies did not focus on a detailed evaluation of atherosclerosis and did not examine all genetic mutations. At this point, our study is the first in the literature and revealed that homozygote mutations are a risk factor for atherosclerosis, regardless of the type of mutation. Thus, this result confirms the information in the literature that homozygote mutations cause worse clinical prognosis.

Çakar et al^[31] found that the severity of the inflammatory state, rather than the attack type and genetic mutations seem to effect the augmented arterial stiffness measurements which is known as another atherosclerotic risk factor. In our study, the inflammatory status was stable and all patients were in attack-free period. Therefore, our study is valuable in terms of showing the parameters associated with atherosclerosis in attack-free FMF patients.

Osteopontin and CRP have collective roles in various inflammatory processes, including atherosclerosis. Increased levels of osteopontin and CRP in atherosclerotic plaques were determined.^[32] It was found that increased osteopontin levels were an idependent risk factor for cardiovasculary disease.^[33] Osteopontin levels were found to be higher in atherosclerosis-related diseases such as hypertension, diabetes mellitus, obesity and chronic kidney disease.^[32,34] To date, osteopontin level and its relationship with atherosclerosis have never been evaluated in FMF patients. Although the patients in our study did not have any comorbidity that may be associated with atherosclerosis, we found a significant positive correlation between osteopontin and CIMT, PAI, AC, and Castelli risk index I. These results are an important indicator that FMF patients are at risk for atherosclerosis, even if they are in an attack-free period. There are conflicting data in the literature regarding the relationship between osteopontin and dyslipidemia. While some studies revealed a relationship between osteopontin and the presence of dyslipidemia, some could not show this relationship.^[35] Contrary to these conflicting results, present study found a significant negative correlation between osteopontin and HDLc. Another study supported our resuts and reported a decrease in osteopontin levels with statin treatment.^[36]

Vitamin D is known to inhibit the vascular calcification, ameliorate endothelial functions, and act against atherosclerosis.^[37] A study reported that cardiovascular mortality was higher in patients with coronary artery disease with low vitamin D levels.^[38] Other study showed the correlation between vitamin D deficiency and arterial stiffness and endothelial dysfunction.^[15] In FMF patients, it was reported that vitamin D levels were lower compared to controls and low vitamin D levels could also trigger the FMF attacks.^[39–41] Although we found that vitamin D levels were low in FMF patients and healthy subjects, there was no significant difference in vitamin D levels between groups. Furthermore, there was also no correlation between vitamin D levels and osteopontin, CIMT and any atherosclerotic indices. A study showed no significant difference of CIMT among FMF patients and controls, however, significant decrease of vitamin D in FMF patients was observed.^[41] Similar to our results, there was also no correlation between vitamin D levels and CIMT in this study. Because of the discrepant results, further prospective studies on the effect of vitamin D on the development of atherosclerosis in FMF patients are required.

The main limitation of this study is relatively small sample size to represent the FMF patients and cardiovascular risk. The other limitation is the cross-sectional design which may be a problem to generalize our results to long term complications. Furthermore, our results need to be extended to a broader FMF population with patients who are in attack period. In our study, we didn’t collect information about diet and physical activity level which can have a confounding effect on the relationship between the evaluated parameters in FMF.

5. Conclusion

FMF patients had higher serum osteopontin level, CIMT, Castelli risk index I, AC, and PAI that have been associated with increased cardiovascular risk, even though the patients were on colchicine treatment during the attack-free period. These results were suggested that increased osteopontin level can be considered supplementary biomarker in the prediction of atherosclerosis in FMF patients. Since there is no study investigating this issue to date, the available data in the literature do not allow for a precise definition of the role of osteopontin in inflammation-induced atherosclerosis in FMF patients. Furthermore, patients who had homozygote mutations had worse atherogenic indices than those with heterozygote mutations.

Author contributions

Conceptualization: Osman Baspinar, Derya Kocer, Oguzhan Sitki Dizdar.

Data curation: Osman Baspinar, Derya Kocer, Turgut Tursem Tokmak.

Formal analysis: Osman Baspinar, Derya Kocer.

Investigation: Derya Kocer, Aslihan Kiraz, Turgut Tursem Tokmak.

Methodology: Derya Kocer, Aslihan Kiraz, Turgut Tursem Tokmak.

Supervision: Oguzhan Sitki Dizdar.

Writing– original draft: Oguzhan Sitki Dizdar.