- Top of page
- Patients and methods
Total Protein S (tPS) and free Protein S (fPS) antigen levels were measured in 3788 healthy blood donors. Men had higher levels of both parameters than women (P < 0·001). Age had no effect on tPS in men, although there was a slight reduction in fPS levels with increasing age. In women increasing age was associated with a significant increase in tPS levels (P < 0·001) but had no effect on fPS after adjustment for menopausal state. Oral contraceptive pill (OCP) use significantly lowered tPS but had no effect on fPS. In post-menopausal women, hormone replacement therapy (HRT) use had no statistically significant effect on either tPS or fPS. Donors with tPS or fPS levels in the lowest percentile (n = 56) were retested; only nine with repeat low levels were identified, eight of whom had persistently low levels over a 4–7-year follow-up. Acquired deficiency was excluded. When possible, family studies were performed, leading to an estimate of prevalence of familial PS deficiency of between 0·03% and 0·13% in the general population.
Protein S (PS), one of the Vitamin K-dependent natural anticoagulants, was first identified in 1977 (Di Scipio et al, 1977). In the blood approximately 60% is in an inactive form bound to C4-binding protein (C4bBP). The rest circulates free in the plasma where it acts as a cofactor for activated Protein C (APC) (Walker, 1980). PS deficiency was first described in individuals with recurrent venous thromboembolism (Comp & Esmon, 1984; Schwartz et al, 1984). It has been identified in 1–7% of patients with deep vein thrombosis (DVT) (Broekmans et al, 1986; Gladson et al, 1988; Ben Tal et al, 1989; Heijboer et al, 1990; Pabinger et al, 1992) but the prevalence of heritable Protein S deficiency in the general population remains unknown. There are inter- and intraindividual variations in PS levels (Melzi d'Ergil et al, 1997) and overlap between normal individuals and those with molecularly defined heterozygous deficiency (Zoller et al, 1995; Simmonds et al, 1997). Levels may also be affected by polymorphisms Protein S Heerlen and Pro 626 in the PS gene (Duchemin et al, 1995; Leroy-Matheron et al, 2000). Free PS (fPS) levels are affected by changes in total PS (tPS) and by variations in the amount of C4bBP available to bind to it. Acquired PS deficiency is not uncommon and can be caused by liver disease or viral infections (D'Angelo et al, 1988; Toulon et al, 1990; Prince et al, 1995; MacCallum et al, 1998).
Women have lower PS levels than men and, although age is not an important determinant of PS level in men, in women increasing age has been associated with rising levels of tPS antigen (Gari et al, 1994; Henkens et al, 1995). Total PS and fPS antigen levels fall during pregnancy (Comp et al, 1986; Fernandez et al, 1989; Clark et al, 1998). The oral contraceptive pill (OCP) causes a reduction in tPS, but no effect on fPS has been reported in many studies (Huisveld et al, 1987; Faioni et al, 1997; Liberti et al, 1999). Recently, however, third generation OCPs have been shown to cause a reduction in fPS levels (Tans et al, 2000). Hormone replacement therapy (HRT) causes a modest, though not statistically significant, reduction in PS activity (Lowe et al, 1997).
Measurement of tPS levels alone, reliance on a single blood sample and the use of universal reference ranges, all of which may be misleading when investigating different patient populations, hamper accurate diagnosis of PS deficiency. A molecular abnormality is identified in only 50–70% of people with apparently low plasma PS levels (Reitsma et al, 1994; Gandrille et al, 1995; Gomez et al, 1995), indicating that the current phenotypic definition of PS deficiency is inaccurate.
The true prevalence of PS deficiency remains unknown and estimates vary widely. Extrapolating from the incidence of PS deficiency in patients with DVT, the prevalence in the community has been estimated at 1 in 33 000 (Gladson et al, 1988). Other estimates based on data from control populations in case–control studies put the prevalence at 1/43–1/143 (Koster et al, 1995; Faioni et al, 1997). In contrast to anti-thrombin and Protein C deficiencies, no studies have examined PS levels in large numbers of healthy people.
We studied Protein S antigen levels in nearly 4000 healthy blood donors with particular reference to age, sex, use of OCPs or HRT and cigarette smoking to produce accurate reference ranges. In addition, prevalence of PS deficiency has been assessed by repeat testing and family studies in individuals with persistently low levels.
- Top of page
- Patients and methods
Both tPS and fPS antigen measurements were available for 3788 healthy blood donors. The population had a mean age of 33 years (32 years for women, 35 years for men) and was made up of 2106 men and 1682 women. The majority of women (1467, 87%) were premenopausal, with 31% of them taking the OCP. The type of OCP was not specified. One hundred and ninety-seven women were post-menopausal and 15% of those were taking HRT. Eighteen women did not specify their menopausal state. Demographic data is shown in Table I.
Table I. Demographic data of 3788 healthy blood donors.
| ||Men (%)||Women (%)|
|Number||2106 (55·6)||1682 (44·4)|
| No||1475 (70)||1244 (74)|
| Yes||427 (30)||423 (26)|
| < 25||534 (25·4)||658 (39·1)|
| 25- < 45||1153 (54·6)||775 (46·1)|
| ≥ 45||419 (20)||249 (14·8)|
|OCP||–||453 (31 of pre)|
|HRT||–||29 (15 of post)|
Summary statistics are shown in Table II. Sex was significantly correlated with PS levels, with men showing higher mean tPS and fPS levels than women (P < 0·001). The difference in adjusted means was 13·6 (CI 12·1–15·1) for tPS and 17·9 (CI 16·1–19·6) for fPS. Increasing age in men showed a non-significant trend to increasing the mean tPS level (P = 0·056), although this was associated with a reduction in mean fPS level (P = 0·001).
Table II. Summary statistics for total PS (tPS) and free PS (fPS).
| || Number||Mean PS(%)|| (95% CI)||Difference in adjusted mean (CI)|| P-value|
|Men||2106||111||(110, 112)||13·6 (12·1, 15·1)||< 0·001|
|Women||1682||97||(95, 98)||–|| |
| Non-OCP||1014||99||(98, 100)||7·41 (5·05, 9·77)||< 0·001|
| OCP||453||91||(89, 92)||–|| |
| Non-HRT||168||101||(96, 106·5)||6·86 (−1·83, 15·55)||0·120|
| HRT||29||95||(85, 104)||–|| |
| Pre-||1467||96||(95, 97)||−0·35 (−4·35, 3·65)||0·864|
| Post-||197||101||(98, 104)||–|| |
|Men||2106||118||(117, 119)||17·9 (16·1, 19·6)||< 0·001|
|Women||1682||100||(98, 101)||–|| |
| Non-OCP||1014||100||(99, 102)||2·55 (−0·21, 5·30)||0·070|
| OCP||453||98||(95, 100)||–|| |
| Non-HRT||168||104||(97, 109)||6·31 (−3·5, 16·18)||0·200|
| HRT||29||97||(86, 108)||–|| |
| Pre-||1467||99||(98, 100)||−1·76 (−6·44, 2·91)||0·460|
| Post-||197||103||(99, 106)||–|| |
In women, multivariate analysis included OCP use and menopausal state in addition to age and cigarette smoking. After adjustment for other variables, increasing age was associated with increasing tPS levels (P < 0·001), whereas menopausal state had no effect on tPS (P = 0·86) after adjusting for age; this indicates that age is a more important variable than menopausal state. OCP use was associated with a lowering of the mean tPS by 8% (P < 0·001). Free PS levels increased with advancing age (P = 0·02) on univariate analysis, although this lost significance on multivariate analysis (P = 0·34), the strongest effect being adjustment for menopausal state. OCP use produced a minimal lowering of the mean fPS (2%, P = 0·07). HRT use within the post-menopausal group of women did not show any effect on tPS (P = 0·12) or fPS levels (P = 0·20).
Cigarette smoking had no effect on tPS levels (P = 0·32). The effect of cigarette smoking on fPS produced an inverted U in the means with increasing number of cigarettes smoked. This was not significant in either group alone (women P = 0·28, men P = 0·19), but when men and women were considered together significance was reached (P = 0·04) owing to the larger sample size.
Table III shows the medians and 2·5–97·5 percentile reference ranges for all relevant groups. Lower limits for tPS varied from 56% to 73% ,the lowest being in young women OCP users, the highest in men. For fPS the lower limit ranged from 51% in young women to 71% in men.
Table III. Age-specific reference ranges for total (tPS) and free (fPS) Protein S levels, showing median and 2·5–97·5 percentiles.
| || Number||Age (years) < 25|| 25 – < 45|| ≥ 45|| Overall|
|Men||2106||106 (72–168)||109 (72–162)||110 (73–171)||108 (72–164)|
| Hormone||487||84 (56–136)||89 (56–148)||95 (57–157)||87 (57–142)|
| Non-hormone||1195||96 (65–141)||96 (67–152)||99 (71–146)||96 (68–148)|
|Men||2106||119 (71–177)||115 (68–177)||113 (67–175)||115 (68–176)|
|Women||1682||96 (51–155)||98 (55–156)||100 (60–152)||98 (54–155)|
On the initial sample, 58 donors had tPS and/or fPS results below our chosen sex-specific cut-off points (tPS: men 62%, women 56%; fPS men 57%, women 47%), corresponding to the lowest 1% of results. There were 19 with low tPS, 33 with low fPS and six with low levels of both. All had normal levels of anti-thrombin and Protein C activity on at least two occasions. Two were lost to follow-up; the remaining 56 donors were retested at least 6 months after the initial result revealing 9/56 (17%) with persistently low levels, either fPS alone or both tPS and fPS, and all had normal liver function and prothrombin times. All those with an initial isolated low tPS had normal results on repeat testing and, within this group, there was an excess of women (13/19), the majority of whom were OCP users (10/13).
Of the nine donors with confirmed low PS levels, one subsequently had normal results and was not studied further; the others, two men and six women, were monitored for a further 4–7 years. Three had reduced tPS and fPS and five showed fPS deficiency only. None has had an episode of venous thrombosis. Family studies were performed and results of these are shown in Table IV. Two have a family history of DVT, the affected family member in one case having normal PS levels. In the other case a maternal uncle had a DVT, although the mother had normal PS levels. In three cases we were unable to obtain samples on any family members. Three had extensive family testing, including either both parents or one parent, all siblings and all children, and showed no evidence of PS deficiency in family members. Two had incomplete family testing in which one had normal relatives and one had a non-identical twin with a low PS level.
Table IV. Details of donors with low Protein S (PS) levels and their families.
| || || || ||Diagnosis PS (%)||Mean PS** follow-up (%)||Family||Relatives|
| || || || || || || || || || |
|1||23||F||Y*||46·5||45·5||45·8||43·1||None||F, 3/3sibs, 2/2 children: N|
|2||27||F||N||45·5||42||60·6||46·2||Maternal uncle DVT||F, M, 2/4 sibs: N|
|3||19||F||Y*||50·5||46·5||46·2||43·3||None||M, 2 sibs: N|
|6||41||M||–||63||52||63·7||49||Mother DVT||F, M, 4/7 sibs, 2/2 children:N|
|8||41||F||N||61||43·5||54·8||47·6||None||1/4 sibs: N, 1/4 sibs low fPS|
Thus, only one case of familial PS deficiency was identified (donor 8), giving a prevalence of 1/3788 (0·026%) (95% CI 0–0·06). However, if the estimate of prevalence includes donors in whom familial PS deficiency cannot be excluded (donors 3, 4, 5, 7), owing to incomplete family studies, it may be higher at 5/3788 (0·13%) (95% CI 0·06–0·24).
- Top of page
- Patients and methods
In a study of 3788 blood donors we found several important factors affecting PS levels. Our findings that women showed lower mean fPS and tPS levels than men and that HRT use had no significant effect on tPS or fPS confirm results of previous smaller studies (Gari et al, 1994; Henkens et al, 1995; Lowe et al, 1997). Although third generation OCPs have been reported to significantly lower both tPS and fPS levels (Tans et al, 2000), our study found an OCP effect only for tPS. Individual donor OCP preparations were not recorded in our study. However, local statistics indicate that around one third of OCP users at that time were taking third-generation preparations (Primary Care Information Statistics Unit, 2001). Hence, any unique effect of third-generation preparations could be diluted by the presence of other OCP preparations.
There was no age effect on tPS in men and the reduction of fPS in older men that has also occurred in other studies (Henkens et al, 1995; Liberti et al, 1999) is probably not clinically significant, having little effect on the lower limit of the 95% reference interval. In women, tPS levels rose with increasing age and this was not caused by menopausal state. This is at variance with results from the smaller Leiden group study (Liberti et al, 1999), which found menopause to be the more important factor. Close linkage between menopause and age may explain these discordant findings. In our study, the effect of age on fPS was lost on multivariate analysis owing to the effect of menopausal state (an age-related parameter), although when age and menopausal state were directly compared there was a trend towards the age effect being stronger than menopause.
Our study shows the wide range of lower limits (2·5% percentile) obtained from different subpopulations (tPS: 56-73%, fPS: 51-71%). These results may be a useful guide to producing population-targeted, sex-based reference ranges. For men a single reference range for each of tPS and fPS would be adequate. For women tPS ranges should include age and OCP use, for fPS only age need be used.
Estimates of community prevalence of PS deficiency have ranged from 1/43 to 1/33,000, this latter by extrapolation from prevalence in patients with DVT: a method that was proved inaccurate when applied to anti-thrombin and Protein C deficiency (Tait et al, 1994, 1995). We used repeat testing of all individuals with results below the first centile and long-term follow-up of those with persistently low results, a method successfully used for other coagulation factors as confirmed by molecular testing (Tait et al, 1994, 1995). Persistently low PS levels in three donors with no evidence of inheritance may be explained by the presence of PS dimorphism in the proband (Duchemin et al, 1995; Leroy-Matheron et al, 2000), although molecular testing was not performed to confirm this.
The prevalence of PS deficiency estimated from this study is between 0·03% and 0·13%. This study may have failed to identify any true deficient individual whose level lies between the 1% and the 2·5% cut-off points. However, this number is probably small, as in other thrombophilic defects in which 12/14 (Protein C) and 11/14 (anti-thrombin) molecularly proven deficient donors had initial levels in the lowest 1 percentile (Tait et al, 1994, 1995). This estimate of prevalence of heterozygous PS deficiency (approximately 1/2000) would predict a homozygous/double heterozygous prevalence of around 1/16 million – in keeping with the reported rarity of this severe condition (Gomez et al, 1994).
In conclusion, in the largest study of the literature, we have looked for factors affecting Protein S levels and found that they vary with sex, age and OCP use. The use of targeted normal ranges should allow more accurate definition of true PS deficiency (Table III). We have identified a number of healthy blood donors with persistently low PS levels and from this information estimated familial PS deficiency to affect 0·03–0·13% of the general population.