/oornal o/ Interrinl Medicine 1991 : 229: 51 7-520
AIMNIS
0954682091 000982
The effect of dietary vitamin K on warfarin-induced anticoagulation F. M. PEDEKSEN, 0. H A M B E R G . K. HESS & 1,. O V E S E N From the Departmerit of lnlcrnal Medicine 111. Copenhagen Munir:ilml Hospitul. Coprnliagrri. Ikrirriurk
Abstract. Pedersen FM. Hamberg 0.Hess K. Ovesen I, (Department of Internal Medicine 111. Copenhagen Municipal Hospital. Copenhagen, Denmark). The effect of dietary vitamin K on warfarin-induced anticoagulation. /ournu/ o/ lntcrnul Medicine 1991 : 229: 5 1 7-520. We examined the effect of vitamin-K-rich vegetables, vitamin-K-poor vegetables and phytomenadione on the stability of warfarin-induced anticoagulation. htients on stable anticoagulant treatment were randomized to either 1 ( n = 5), 2 ( t i = 7) or 7 ( n = 13) d with high intake of vitamin-K-rich vegetables (median daily vitamin K intake I 1 0 0 jig) or high intake of vitamin-K-poor vegetables (daily vitamin K intake 135 p g ) for 6 d ( n = 7). or habitual diet supplemented with 1000 j i g of phytornenadione daily ( n = 5). Nine patients (69%: 95% CI. 39-91 %) who consumcd vitamin-K-rich vegetables for 7 d reached activities above the therapeutic level. Two (40%;953, CI. 5-850/,) and three patients (43%:95% CI. 10-860/,) who consumed vitamin-K-rich vegetables for 1 and 2 d. respectively, exceeded the upper therapeutic limit. No changes were observed in the vitamin-K-poor group. All patients who received phytomenadionc exceeded the upper therapeutic limit. Dietary vitamin K should be regarded as a n important environmental factor contributing to unwanted disturbances in warfarin-induced anticoagulation. Ke!gwords:
anticoagulation. vitamin K. warfarin.
Introduction Keports on acquired warfarin resistance due to ingestion of large amounts of vitamin-K-rich vegetables suggest that dietary vitamin K is a potential environmental factor contributing to unstable anticoagulation [l]. In this study we examined the effect on stability of warfarin-induced anticoagulation of consumption of vitamin-K-rich vegetables for 1, 2 , and 7 d, as well as the effect of equivalent amounts of oral phytomenadione. and consumption of vitamin-K-poor vegetables.
Patients and methods A total of 37 out-patients on stable anticoagulation for a period of a t least 3 weeks prior to the study with a fixed dose of warfarin participated in the study. Patients included in protocols 1 and 5 (see below) had been treated for 3 months a n d were about to stop anticoagulant therapy.
Plasma coagulant activity was measured by the prothrornbin-proconvertin method with a commercial reagent (Normotest). The therapeutic range was 10-25%, corresponding to 3.6-2.0 in terms of the International Normalised Ratio 1.21. All patients gave their informed consent before entering the study. The protocols were approved by the local ethics committee.
Protocols Protocol 1 was performcd prior to the other protocols. Patients were randomly allocated to one of protocols 2-5. To confirm the stability of anticoagulation, all protocols included a n initial control period of 5 d. during which plasma coagulant activity was measured on alternate days. The intervention period was followed by a second control period in order to study changes in plasma coagulant activity after cessation of vitamin K intervention. The following choice of vitamin-K-rich vegetables
51 7
518
F. M. PEDEKSEN et nl.
Table 1. Percentage plasma coagulant activity (Normotest) in patients on stable anticoagulant therapy beforc. during and after dictary intcrvcntion with vitamin-K-rich vcgctables for 7 d (protocol I ) . 1 d (protocol 2 ) . 2 d (protocol 3). vegetables poor in vitamin K for 6 d (protocol 4 ) and phytomenadionc 1 mg daily for 6 d (protocol 5) Protocol 1 (ri = 1 3 ) 7 d with vitamin K Day 0 31 51 71 lo 12
20 23 27 31 27 27
( 1 1-24) (14-49) (17-63). (18-55)* (IO-~I)* (13-45)*
Protocol 2 (ri = 5) 1 d with vitamin K
Protocol 3 (ri = 7) 2 d with vitamin K
Day
Ihy 0 11 21 3 4 7
0
16
II 2 3 4 7
I6 18
21 23 18
(11-22) (9-24) (11-29) (14-34)' (15-33)' (14-32)
Protocol 4 (n = 7) 6 d with low vitamin K
Protocol 5 (n = 5) 6 d with phytomenadione ( 1 nig)
Day 0 II 21 31 41 7 9 11
Day 0
19 18
20 19 18 18
19 21
(11-23) (12-24) (11-25) (11-24) (9-25) (9-24) (9-25) (9-26)
11
21 31 41 7 9 11
21 21 28 46 56 72 66 65
17 16 18
21 20 IX
(11-21) (11-24) (13-28) ( 1 7-36)' (19-34)' (15-34)'
(13-25) (16-36) (22-50) (27-XU) (31-74)* (50-110)' (52-83)' (43-80)*
The results arc expressed as median values, with ranges in parentheses. I = days of intervention. Thc therapeutic rangc w a s 10-25'%,. *lknotes statistically significant difference ( P < 0.05) compared to day 0.
was offered in protocols 1, 2 and 3 : 300 g of spinach or 4 0 0 g of broccoli or 4 0 0 g of brussels sprouts or 750 g of lettuce d-', giving c. 1000 j i g of vitamin K per day [3,4]. Protocol 1. (n = 13) (four women and nine men. age range 20-72 years). Seven days with high intake of vitamin-K-rich vegetables. Plasma coagulant activity was measured on days 3, 5, 7. 10 and 12. Protocol 2. ( n = 5) (two women and three men. age range 27-73 years). One day with high intake of vitamin-K-rich vegetables. Plasma coagulant activity was measured on days 1. 2, 3. 4 and 7. Protocol 3. (n = 7) (three women and four men. age range 31-77 years). Two days with high intake of vitamin-K-rich vegetables. Plasma coagulant activity was measured on days 1. 2, 3 . 4 and 7. Protocol 4. (n = 7) (two women and five men, age range 52-74 years). Seven days with high intake of vitamin-K-poor vegetables (tomatoes, cucumber. potatoes or onion, to give a total of 500 g d-') Plasma coagulant activity was measured on days 1, 2, 3 , 4, 7, 9 and 11.
Protocol 5. (11 = 5) (one woman and four men, age range 64-78 years). Six days with supplementation of 1000 p g of oral phytomenadione (Konakion, Roche A/S, Denmark). Plasma coagulant activity was measured on days 1. 2, 3. 4, 7. 9 and 11. Daily food and drink intake was recorded prior to and during the intervention periods. Median habitual intake of vitamin K was 190 p g d-' (range 50-280 p g d-') and was increased to 1100 p g d-' (range 600-1 500 pg d-') in protocols 1-3, and to 1 2 0 0 j i g d-' (range 1100-1230 p g d-I) in protocol 5. Vitamin K intake was reduced to 1 3 5 p g d-' (range 125-200 pg d-') in protocol 4. Statistical analysis The results are expressed as median values, with ranges shown in parentheses. Statistical differences between plasma coagulant activities on different days for each protocol were examined by the Friedman two-way analysis of variance. If P-values of < 0.05 were obtained, further multiple comparisons with day 0 (last day in first control period) were performed [S].
ANTICOAGULATION AND VITAMIN K
6o
t I
I
I control
I
2
I control
control
p c )
I
2 control
C ( d ) 40
40
519
95% CI. 39-91%) reached values above the therapeutic range (Table 1 and Fig. 1).These disturbances lasted for several days after resumption of the patients’ habitual diet. Even 1 and 2 d of intake of vitamin-K-rich vegetables was sufficient to increase plasma coagulant activity, yielding values above the therapeutic range in two (40%; 95% CI, 5 4 5 % ) and three ( 43%; 95% CI. 1 0 4 6 % ) patients, respectively. No changes in plasma coagulant activity were observed after eating a diet rich in vitamin-K-poor vegetables for 6 d. whereas supplementation with phytomenadione for 6 d caused a sharp increase in plasma coagulant activity.
Discussion
I control
I
I
I
control
2
I
control
control
I
2 control
2 control
Fig. 1. 1’crcent;igc individual changes in plasma coagulant activity (Normotest) before. during and after dietary intervention with vitamin-K-rich vegetables for (a) 7 d. (b) 1 d. (c) 2 cl. (cl) with vegetables poor in vitamin K. and ( e ) with phytoiiieriadione 1 ing daily for h d. 1 control is the value obtiiincd immediately before intervention. 1 is the highest value reached during the intervention period (a, d and e) or 2-3 d after intervention (b. c). and 2 control is the activity 5-6 d after intervention has ceased. The stippled area indicates the therapeutic range.
Results When patients replaced their habitual diet with a diet rich in vitamin K for 7 d, plasma coagulant activity increased on day 5, and nine of the patients ( 6 9 % ;
This study demonstrates that a high intake of vitamin-K-rich vegetables can interfere with oral anticoagulant therapy. The response varied, but in all vitamin-K-rich protocols some patients reached levels of plasma coagulant activity above the therapeutic range. Therefore dietary vitamin K should be regarded as an important factor contributing to unstable anticoagulation. It seems reasonable to assume that the effect on plasma coagulant activity of the high intake of vegetables was due to their high vitamin-K content. The much more pronounced and consistent effect exerted by oral phytomenadione is probably explained by overestimation of thc vitamin-K content of the vegetables, particularly after cooking. Furthermore, phytomenadione was administered once daily, which could lead to higher intrahepatocytic vitamin-K concentrations, favouring more efficient competition with warfarin. Differences in absorption between phytomenadione and vitamin K in vege‘tables are unlikely, as dietary vitamin-K absorption is reported to be very effective [6]. A study by Karlson et al. [7] found no effect on anticoagulation levels of a single dose of 250 pg of phytomenadione or one day’s intake of 250 g of either broccoli or spinach with an estimated vitaminK content of about 200-8OOpg. However, a continuous intake of this amount of vitamin K caused plasma coagulant activity to increase and exceed the upper therapeutic limit within a few days. The present study suggests that even a single day‘s increase of c. 1000 pg of dietary vitamin K can cause an undesirable increase in plasmii coagulant activity, and that this effect is accentuated by a continuous
52C)
F. M. I’EDEKSEN et nl.
daily intake. causing disturbances that last for several days after resumption of the patients’ habitual diet. Taken together with the results of Karlson et a/., the lower limit of daily variation in dietary vitamin-K intake should not exceed 250-500 pg.
Food Composition ‘I’ables I 9 8 9 . leviidsr,iiddelst!/relscri (Publication No. SC3). 1989. Copenhagen, 1)enmerk. Parrish I). Iktcrmination of vitamin K in food. CKC Crit K c v Food Sci Nutr 19x0: 13: 337-52. Siege1 S. Castellan Nj. Nori-Pnrrirrietric Stlitistius /or tlic BclinviornlSciences. 2nd edn. New York: McGraw Hill Hook Company. 1988: 174-81. Barkham 1’. Shearer MJ. Metabolism of vitamin K. l’roc K S n f : M d 1977: 8 0 : 9 3 4 .
References I Kcinpin SJ. Warfarin resistance causcd by broccoli. N Erigl /
Karlson B. k i j d B. Hcllstrom. On the influence of vitamin-Krich vegetables and wine on the cffcctivcncss of warfarin treatment. Actri Mrd Scrtrrrl 1986; 220: 347-50.
MnI 1 9 8 3 : 3 0 8 : 1229.
2 World H r n l f l i Orgaiiizotiori Exprrt Conirriittw ori Hiologicnl Sfrrrrr(nr[lisnfiori. 33rd K r p r f . Geneva : World Hcelth Organisation. 1983: 81-105 (WHO Technical Report Series 687).
Received 2X May 1990. accepted 2 1 I)cccmbcr 1990. Corresporidericc: Ole Hamberg. MI). Medical Department A-2 I 5 I , Rigshospitalet. 9 Hlegdamsvej. IIK-2 100 Copenhagen. Iknniark.
AIMNIS
0954682091 000982
The effect of dietary vitamin K on warfarin-induced anticoagulation F. M. PEDEKSEN, 0. H A M B E R G . K. HESS & 1,. O V E S E N From the Departmerit of lnlcrnal Medicine 111. Copenhagen Munir:ilml Hospitul. Coprnliagrri. Ikrirriurk
Abstract. Pedersen FM. Hamberg 0.Hess K. Ovesen I, (Department of Internal Medicine 111. Copenhagen Municipal Hospital. Copenhagen, Denmark). The effect of dietary vitamin K on warfarin-induced anticoagulation. /ournu/ o/ lntcrnul Medicine 1991 : 229: 5 1 7-520. We examined the effect of vitamin-K-rich vegetables, vitamin-K-poor vegetables and phytomenadione on the stability of warfarin-induced anticoagulation. htients on stable anticoagulant treatment were randomized to either 1 ( n = 5), 2 ( t i = 7) or 7 ( n = 13) d with high intake of vitamin-K-rich vegetables (median daily vitamin K intake I 1 0 0 jig) or high intake of vitamin-K-poor vegetables (daily vitamin K intake 135 p g ) for 6 d ( n = 7). or habitual diet supplemented with 1000 j i g of phytornenadione daily ( n = 5). Nine patients (69%: 95% CI. 39-91 %) who consumcd vitamin-K-rich vegetables for 7 d reached activities above the therapeutic level. Two (40%;953, CI. 5-850/,) and three patients (43%:95% CI. 10-860/,) who consumed vitamin-K-rich vegetables for 1 and 2 d. respectively, exceeded the upper therapeutic limit. No changes were observed in the vitamin-K-poor group. All patients who received phytomenadionc exceeded the upper therapeutic limit. Dietary vitamin K should be regarded as a n important environmental factor contributing to unwanted disturbances in warfarin-induced anticoagulation. Ke!gwords:
anticoagulation. vitamin K. warfarin.
Introduction Keports on acquired warfarin resistance due to ingestion of large amounts of vitamin-K-rich vegetables suggest that dietary vitamin K is a potential environmental factor contributing to unstable anticoagulation [l]. In this study we examined the effect on stability of warfarin-induced anticoagulation of consumption of vitamin-K-rich vegetables for 1, 2 , and 7 d, as well as the effect of equivalent amounts of oral phytomenadione. and consumption of vitamin-K-poor vegetables.
Patients and methods A total of 37 out-patients on stable anticoagulation for a period of a t least 3 weeks prior to the study with a fixed dose of warfarin participated in the study. Patients included in protocols 1 and 5 (see below) had been treated for 3 months a n d were about to stop anticoagulant therapy.
Plasma coagulant activity was measured by the prothrornbin-proconvertin method with a commercial reagent (Normotest). The therapeutic range was 10-25%, corresponding to 3.6-2.0 in terms of the International Normalised Ratio 1.21. All patients gave their informed consent before entering the study. The protocols were approved by the local ethics committee.
Protocols Protocol 1 was performcd prior to the other protocols. Patients were randomly allocated to one of protocols 2-5. To confirm the stability of anticoagulation, all protocols included a n initial control period of 5 d. during which plasma coagulant activity was measured on alternate days. The intervention period was followed by a second control period in order to study changes in plasma coagulant activity after cessation of vitamin K intervention. The following choice of vitamin-K-rich vegetables
51 7
518
F. M. PEDEKSEN et nl.
Table 1. Percentage plasma coagulant activity (Normotest) in patients on stable anticoagulant therapy beforc. during and after dictary intcrvcntion with vitamin-K-rich vcgctables for 7 d (protocol I ) . 1 d (protocol 2 ) . 2 d (protocol 3). vegetables poor in vitamin K for 6 d (protocol 4 ) and phytomenadionc 1 mg daily for 6 d (protocol 5) Protocol 1 (ri = 1 3 ) 7 d with vitamin K Day 0 31 51 71 lo 12
20 23 27 31 27 27
( 1 1-24) (14-49) (17-63). (18-55)* (IO-~I)* (13-45)*
Protocol 2 (ri = 5) 1 d with vitamin K
Protocol 3 (ri = 7) 2 d with vitamin K
Day
Ihy 0 11 21 3 4 7
0
16
II 2 3 4 7
I6 18
21 23 18
(11-22) (9-24) (11-29) (14-34)' (15-33)' (14-32)
Protocol 4 (n = 7) 6 d with low vitamin K
Protocol 5 (n = 5) 6 d with phytomenadione ( 1 nig)
Day 0 II 21 31 41 7 9 11
Day 0
19 18
20 19 18 18
19 21
(11-23) (12-24) (11-25) (11-24) (9-25) (9-24) (9-25) (9-26)
11
21 31 41 7 9 11
21 21 28 46 56 72 66 65
17 16 18
21 20 IX
(11-21) (11-24) (13-28) ( 1 7-36)' (19-34)' (15-34)'
(13-25) (16-36) (22-50) (27-XU) (31-74)* (50-110)' (52-83)' (43-80)*
The results arc expressed as median values, with ranges in parentheses. I = days of intervention. Thc therapeutic rangc w a s 10-25'%,. *lknotes statistically significant difference ( P < 0.05) compared to day 0.
was offered in protocols 1, 2 and 3 : 300 g of spinach or 4 0 0 g of broccoli or 4 0 0 g of brussels sprouts or 750 g of lettuce d-', giving c. 1000 j i g of vitamin K per day [3,4]. Protocol 1. (n = 13) (four women and nine men. age range 20-72 years). Seven days with high intake of vitamin-K-rich vegetables. Plasma coagulant activity was measured on days 3, 5, 7. 10 and 12. Protocol 2. ( n = 5) (two women and three men. age range 27-73 years). One day with high intake of vitamin-K-rich vegetables. Plasma coagulant activity was measured on days 1. 2, 3. 4 and 7. Protocol 3. (n = 7) (three women and four men. age range 31-77 years). Two days with high intake of vitamin-K-rich vegetables. Plasma coagulant activity was measured on days 1. 2, 3 . 4 and 7. Protocol 4. (n = 7) (two women and five men, age range 52-74 years). Seven days with high intake of vitamin-K-poor vegetables (tomatoes, cucumber. potatoes or onion, to give a total of 500 g d-') Plasma coagulant activity was measured on days 1, 2, 3 , 4, 7, 9 and 11.
Protocol 5. (11 = 5) (one woman and four men, age range 64-78 years). Six days with supplementation of 1000 p g of oral phytomenadione (Konakion, Roche A/S, Denmark). Plasma coagulant activity was measured on days 1. 2, 3. 4, 7. 9 and 11. Daily food and drink intake was recorded prior to and during the intervention periods. Median habitual intake of vitamin K was 190 p g d-' (range 50-280 p g d-') and was increased to 1100 p g d-' (range 600-1 500 pg d-') in protocols 1-3, and to 1 2 0 0 j i g d-' (range 1100-1230 p g d-I) in protocol 5. Vitamin K intake was reduced to 1 3 5 p g d-' (range 125-200 pg d-') in protocol 4. Statistical analysis The results are expressed as median values, with ranges shown in parentheses. Statistical differences between plasma coagulant activities on different days for each protocol were examined by the Friedman two-way analysis of variance. If P-values of < 0.05 were obtained, further multiple comparisons with day 0 (last day in first control period) were performed [S].
ANTICOAGULATION AND VITAMIN K
6o
t I
I
I control
I
2
I control
control
p c )
I
2 control
C ( d ) 40
40
519
95% CI. 39-91%) reached values above the therapeutic range (Table 1 and Fig. 1).These disturbances lasted for several days after resumption of the patients’ habitual diet. Even 1 and 2 d of intake of vitamin-K-rich vegetables was sufficient to increase plasma coagulant activity, yielding values above the therapeutic range in two (40%; 95% CI, 5 4 5 % ) and three ( 43%; 95% CI. 1 0 4 6 % ) patients, respectively. No changes in plasma coagulant activity were observed after eating a diet rich in vitamin-K-poor vegetables for 6 d. whereas supplementation with phytomenadione for 6 d caused a sharp increase in plasma coagulant activity.
Discussion
I control
I
I
I
control
2
I
control
control
I
2 control
2 control
Fig. 1. 1’crcent;igc individual changes in plasma coagulant activity (Normotest) before. during and after dietary intervention with vitamin-K-rich vegetables for (a) 7 d. (b) 1 d. (c) 2 cl. (cl) with vegetables poor in vitamin K. and ( e ) with phytoiiieriadione 1 ing daily for h d. 1 control is the value obtiiincd immediately before intervention. 1 is the highest value reached during the intervention period (a, d and e) or 2-3 d after intervention (b. c). and 2 control is the activity 5-6 d after intervention has ceased. The stippled area indicates the therapeutic range.
Results When patients replaced their habitual diet with a diet rich in vitamin K for 7 d, plasma coagulant activity increased on day 5, and nine of the patients ( 6 9 % ;
This study demonstrates that a high intake of vitamin-K-rich vegetables can interfere with oral anticoagulant therapy. The response varied, but in all vitamin-K-rich protocols some patients reached levels of plasma coagulant activity above the therapeutic range. Therefore dietary vitamin K should be regarded as an important factor contributing to unstable anticoagulation. It seems reasonable to assume that the effect on plasma coagulant activity of the high intake of vegetables was due to their high vitamin-K content. The much more pronounced and consistent effect exerted by oral phytomenadione is probably explained by overestimation of thc vitamin-K content of the vegetables, particularly after cooking. Furthermore, phytomenadione was administered once daily, which could lead to higher intrahepatocytic vitamin-K concentrations, favouring more efficient competition with warfarin. Differences in absorption between phytomenadione and vitamin K in vege‘tables are unlikely, as dietary vitamin-K absorption is reported to be very effective [6]. A study by Karlson et al. [7] found no effect on anticoagulation levels of a single dose of 250 pg of phytomenadione or one day’s intake of 250 g of either broccoli or spinach with an estimated vitaminK content of about 200-8OOpg. However, a continuous intake of this amount of vitamin K caused plasma coagulant activity to increase and exceed the upper therapeutic limit within a few days. The present study suggests that even a single day‘s increase of c. 1000 pg of dietary vitamin K can cause an undesirable increase in plasmii coagulant activity, and that this effect is accentuated by a continuous
52C)
F. M. I’EDEKSEN et nl.
daily intake. causing disturbances that last for several days after resumption of the patients’ habitual diet. Taken together with the results of Karlson et a/., the lower limit of daily variation in dietary vitamin-K intake should not exceed 250-500 pg.
Food Composition ‘I’ables I 9 8 9 . leviidsr,iiddelst!/relscri (Publication No. SC3). 1989. Copenhagen, 1)enmerk. Parrish I). Iktcrmination of vitamin K in food. CKC Crit K c v Food Sci Nutr 19x0: 13: 337-52. Siege1 S. Castellan Nj. Nori-Pnrrirrietric Stlitistius /or tlic BclinviornlSciences. 2nd edn. New York: McGraw Hill Hook Company. 1988: 174-81. Barkham 1’. Shearer MJ. Metabolism of vitamin K. l’roc K S n f : M d 1977: 8 0 : 9 3 4 .
References I Kcinpin SJ. Warfarin resistance causcd by broccoli. N Erigl /
Karlson B. k i j d B. Hcllstrom. On the influence of vitamin-Krich vegetables and wine on the cffcctivcncss of warfarin treatment. Actri Mrd Scrtrrrl 1986; 220: 347-50.
MnI 1 9 8 3 : 3 0 8 : 1229.
2 World H r n l f l i Orgaiiizotiori Exprrt Conirriittw ori Hiologicnl Sfrrrrr(nr[lisnfiori. 33rd K r p r f . Geneva : World Hcelth Organisation. 1983: 81-105 (WHO Technical Report Series 687).
Received 2X May 1990. accepted 2 1 I)cccmbcr 1990. Corresporidericc: Ole Hamberg. MI). Medical Department A-2 I 5 I , Rigshospitalet. 9 Hlegdamsvej. IIK-2 100 Copenhagen. Iknniark.
