Sent March 4, 2007. Feedback by David K. Cundiff, MD
Re: BA Hutten, MH Prins. Duration of treatment with vitamin K antagonists in symptomatic venous thromboembolism. The Cochrane Database of Systematic Reviews 2006:Art. No.: CD001367.pub2. DOI: 10.1002/14651858.CD001367.pub2.
I will comment on aspects of this review according to section.
Background
“……..Patients are usually treated with an initial course of heparin or low-molecular-weight heparin (for approximately 6 days) associated with oral anticoagulant therapy started simultaneously and continued for a period thereafter…..”
Comment: The evidence-basis for this “standard treatment” for venous thromboembolism (VTE) is not referenced. The three small published RCTs comparing heparin and oral anticoagulant therapy with placebo or NSAIDs show no benefit in clinical outcomes with anticoagulants (e.g., mortality from fatal pulmonary emboli (PE) or symptomatic VTE recurrences).1-6
“This prolonged use of oral anticoagulants has proven efficacy in comparison to placebo and low dose heparin (Hull 1979; Lagerstedt 1985)………”
Comment: I will discuss these lines of evidence in sequence.
(1)
(2) Since the Lagerstedt trial did not have a blinded or partly blinded outcome assessment, the authors excluded it from this Cochrane review.
In the Lagerstedt trial of 46 calf vein DVT subjects,8 both groups had an initial course of heparin. DVT recurrences were seen in 0/23 patients treated with three months warfarin versus 8/29 not treated with warfarin. (6/29 were symptomatic). At 90 days 1/23 in the warfarin group had an abnormal perfusion lung scan. Three of 29 subjects not on warfarin had abnormal lung scans, all of whom had recurrent DVT. One of these was symptomatic but not fatal. Again, given the serious bleeding risk of three months of warfarin treatments and the absence of fatal PE in either group, this trial provides no compelling evidence in favor of anticoagulation treatment of VTE.
In contrast to the Lagerstedt trial, a study by Moser and LeMoine found that 0 of 21 patients with no anticoagulant treatment for calf vein DVT had symptomatic PE or silent PE on lung scans, and none developed physical findings or symptoms of postphlebitic syndrome.9
Types of outcome measures
Primary outcome
Incidence of recurrent venous thromboembolism.
Secondary outcomes
(a) Incidence of major bleeding,
(b) Mortality.
Comment: Recurrent DVT
(usually about 90% of all VTE recurrences) is no more and possibly less
clinically serious than major bleeding. Consequently, the primary outcome
should be composite endpoint of recurrent symptomatic VTE and major hemorrhage.
One secondary endpoint of this
review should be “fatal PE,” reflecting that the most important purpose of any
treatment after a diagnosis of VTE is to prevent fatal PE. Other important
secondary endpoints should be fatal bleeding and permanently disabling bleeding
(e.g., nonfatal intracerebral hemorrhage).
Clinicians give VKA to VTE
patients to prevent deaths from PE, not to cure cancer or cardiovascular
disease. Since no more than 10% of deaths in patients with VTE histories are
due to recurrent PE, the overall mortality of this patient population with a
high prevalence of cancer and cardiovascular disease is not relevant to the
efficacy of manipulations of the VKA duration. Fatal PE should be an endpoint.
Overall mortality as an endpoint should be put in the context of the natural
history of VTE.
Results
Incidence
of recurrent VTE
”Six of the eight studies evaluated the incidence of recurrent venous
thromboembolism in the period after cessation of study medication. The
individual studies did not show a statistically significant increase in venous
thromboembolic events among the participants in the long arm after cessation of
treatment. Combining these studies, 96 (7%) of 1304 participants who were
treated for the longer period with vitamin K antagonists, and 78 (6%) of the
1301 participants who were treated a shorter period, experienced a recurrence.
Analysis of the pooled data showed no difference in the incidence of
recurrences during this period (OR 1.24, 95% CI 0.91 to 1.69). Although a
somewhat higher risk was found for participants in the long duration arm after
cessation of treatment compared to those in the short duration arm, a rebound
effect could not be demonstrated.”
Comment: Rebound hypercoagulability (elevated fibinogen levels and activated thromboplastin times) occurs invitro in the first few weeks after discontinuation of VKA and by two months these tests are no longer abnormal.10 Table 1 (below) shows that VTE recurrences in these eight trials are significantly more frequent in the two months following discontinuation of VKA than subsequently (OR: 3.62 (2.79, 4.70)).
Table 1. VTE recurrences within 2 months of stopping VKA versus recurrences thereafter
|
Trial
/ duration / type of VTE |
Rebound
VTE recurrences in the 2 months after stopping VKA / n (% recurrences /
month) |
Subsequent
# of VTE recurrences / n / # months (% recurrences / month) |
OR
(95% CI) |
|
Pinede
/ 12 weeks / P-DVT or PE |
3/268
(0.56) |
16/262
/ 12 / (0.51) |
1.10
(0.32, 3.79) |
|
Pinede
/ 24 weeks P-DVT or PE |
4/264
(0.76) |
14/260
/ 9 / (0.60) |
1.27
(0.41, 3.89) |
|
Agnelli
(2003) / 3 months / PE |
3/161
(0.93) |
15/158
/ 31 (0.31) |
3.04
(0.87, 10.62) |
|
Agnelli
(2003) / 6-12 months / PE |
2/165
(0.61) |
13/163
/ 23 / (0.35) |
1.75
(0.39, 7.81) |
|
Agnelli
(2001) / 3 months / DVT |
1/133
(0.38) |
20/132
/ 35 / (0.44) |
0.87
(0.12, 6.51) |
|
Agnelli
(2001) / 1 year / DVT |
4/134
(1.04) |
16/130
/ 26 / (0.31) |
3.38
(1.12, 10.17) |
|
Kearon
(1999) / 3 months / 1st idiopathic VTE |
5/83
(3.01) |
12/78
/ 22 (0.70) |
4.52
(1.52, 16.51) |
|
Kearon
(1999) / 24 months / 1st idiopathic VTE |
Subjects
not taken off of VKA during the trial. |
|
|
|
Kearon
(2004) / 1 month / 1st VTE transient risk factor |
2/84
(1.19) |
3/82
/ 8 (0.46) |
2.67
(0.44, 6.22) |
|
Kearon
(2004) / 3 months / 1st VTE transient risk factor |
1/81
(0.62) |
1/80
/ 8 / (0.16) |
3.98
(0.25, 64.44) |
|
Levine
/ 4 weeks / P-DVT |
9/105
(4.29) |
3/96
/ 8 / (0.39) |
12.00
(3.19, 5.18) |
|
Levine
/ 3 months / P-DVT |
Timing
of VTE recurrences not noted |
|
|
|
Schulman
(1995) / 6 weeks / 1st VTE |
33/441
(3.74) |
43/408
/ 20.5 / (0.52) |
7.79
(4.89, 12.39) |
|
Schulman
(1995) / 6 months / 1st VTE |
9/451
(1.00) |
28/442
/ 16 / (0.39) |
2.57
(1.21, 5.49) |
|
Schulman
(1997) / 6 months / 2st VTE |
4/111
(1.80) |
19/107
/ 40 / (0.45) |
4.17
(1.40, 2.47) |
|
Schulman
(1997) / indefinite / 2st VTE |
Subjects
not taken off of VKA during the trial. |
||
|
Total |
80/2481
(1.61) |
203/2398
(0.44) |
3.62
(2.79, 4.70) |
The initiation of VKA treatment is also relatively thrombogenic, since the risk of thrombotic events when the INR is less than 1.5 is 7.6 times the risk when the INR is 2.0-2.99 in patients taking warfarin for various indications.11 In an observational study of 2745 continuously anticoagulated patients, Palareti and colleagues showed that the rate of thromboic events in the first 90 days of VKA treatment was almost twice the rate as after the first 90 days (34 recurrences in the first 90 days [686 patient-years] versus 36 VTE recurrences while continuing VKA thereafter (1325 patient-years, OR: 1.82 (1.13, 2.94)).11
Table 2 below shows that, relative to the rate of recurrent VTE after the two month rebound hypercoagulability period, VKA treatment reduces VTE recurrences by 52%. However, Table 3 (below) demonstrates that, when VTE recurrences while subjects are taking VKA are added to rebound recurrences in the two months after discontinuation of VKA (a measure of the overall effect of VKA treatment), the combined monthly rate of recurrence is 43% higher than the subsequent rate of recurrence. This pattern in these eight trials signifies that rebound hypercoagulability causes most of the early VTE recurrences in the groups withdrawn from warfarin and that the hypercoagulable state associated with VKA initiation may also cause some early recurrences.12
Mortality
Comment: About 90% of deaths in
these trials were not due to fatal PE (Table 4 below), confirming that overall
mortality tells us nothing about the efficacy of manipulations of the duration
of VKA treatment.
Even though the most important
potential benefit of VKA treatment for VTE is to prevent fatal PE, using fatal
PE as an endpoint for this or any VTE treatment trial or review has several
problems. Anticoagulation treatment for VTE
trials, including the ones in this review, have a low rate of fatal PE, and
most diagnoses of fatal PE are made on clinical grounds rather than autopsies.
The reported concordance between antemortem diagnoses of PE and confirmation at
autopsy in this era of lung scans and angiograms is still only 10%,13 15%,14 30%,15 and 32%.16 Of those people dying of autopsy confirmed PE, most have
underlying terminal illnesses, leading to prolonged bed rest, inactivity,
venous stasis, and reduced cardiovascular reserve. In patients with postmortem
diagnoses of fatal PE, the proportion with underlying terminal illnesses has
been reported as 95% (169/178)14 and 96.5% (1867/1934).13 Consequently, fatal PE does not necessarily shorten life
expectancy.
Disregarding these caveats about the utility of the endpoint fatal PE for a moment, Table 4 shows the 12 confirmed or suspected deaths from PE. Two of the three PE deaths, for which the timing was reported, occurred within two months of discontinuing the VKA (i.e., likely caused by rebound hypercoagulability). Four of the 12 patients dying of confirmed or suspected PE also had cancer. Other medical diagnoses of other eight patients dying of PE were not reported. From what was reported in these eight trials, most or all of the 12 patients possibly dying of PE may have died while taking VKA (timing not reported for most PE deaths), of rebound hypercoagulability due to VKA, or of their underlying terminal diseases.
Table 2. VTE recurrences while on VKA versus recurrences after two months past discontinuation of VKA
|
Trial
/ duration / type of VTE |
VTE
recurrences while on VKA / n / # months (% recurrences / month) |
VTE
recurrences after two months past stopping VKA / n / # months (% recurrences
/ month) |
OR
(95% CI) |
|
Pinede
/ 12 weeks / P-DVT or PE |
2/268
/ 3 (0.25) |
16/262
/ 12 (0.51) |
0.49
(0.11, 2.13) |
|
Pinede
/ 24 weeks P-DVT or PE |
5/264
/ 6 (0.32) |
14/260
/ 9 (0.60) |
0.53
(0.19, 1.47) |
|
Agnelli
(2003) / 3 months / PE |
No
VKA given after trial began |
||
|
Agnelli
(2003) / 6-12 months (3-9 months on study) / PE |
1/165
/ 6.3 (0.10) |
13/163
/ 23 (0.35) |
0.28
(0.04, 2.13) |
|
Agnelli
(2001) / 3 months / DVT |
No
VKA given after trial began |
||
|
Agnelli
(2001) / 1 year / DVT |
1/134
/ 12 (0.06) |
16/130
/ 26 (0.31) |
0.13
(0.02, 0.99) |
|
Kearon
(1999) / 3 months / 1st idiopathic VTE |
No
VKA given after trial began |
||
|
Kearon
(1999) / 24 months / 1st idiopathic VTE |
Subjects
not taken off of VKA during the trial. |
||
|
Kearon
(2004) / 1 month / 1st VTE transient risk factor |
No
VKA given after trial began |
||
|
Kearon
(2004) / 3 months / 1st VTE transient risk factor |
1/81
/ 2 (0.62) |
1/80
/ 8 (0.16) |
3.95
(0.25,3.50) |
|
Levine
/ 4 weeks / P-DVT |
No
VKA given after trial began |
||
|
Levine
/ 3 months / P-DVT |
Timing
of recurrences not available |
||
|
Schulman
(1995) / 6 weeks / 1st VTE |
4/441/
1.5 (0.60) |
43/408
/ 20.5 (0.52) |
1.18
(0.42, 3.28) |
|
Schulman
(1995) / 6 months / 1st VTE |
6/451
/ 6 (0.22) |
28/442
/ 16 (0.39) |
0.56
(0.23, 1.35) |
|
Schulman
(1997) / 6 months / 2st VTE |
0/111
/ 6 (0.0) |
19/107
/ 40 (0.45) |
0.17
(0.01, 2.81) |
|
Schulman
(1997) / indefinite / 2st VTE |
Subjects
not taken off of VKA during the trial. |
||
|
Total |
20/2481
(0.22) |
150/2398
(0.45) |
0.48
(0.30, 0.77) |
Table 3. VTE recurrences while on VKA and within 2 months of stopping VKA versus recurrences thereafter
|
Trial
/ duration / type of VTE |
Total
recurrences while on VKA and until 2 months past stopping VKA /n / # months
f/u (% recurrences / month) |
Subsequent
# of VTE recurrences / n / # months f/u (% recurrences / month) |
OR
(95% CI) |
|
Pinede
/ 12 weeks / P-DVT or PE |
5/268
/ 5 (0.37) |
16/260
/ 12 / (0.51) |
0.98 (0.35,
2.80) |
|
Pinede
/ 24 weeks P-DVT or PE |
9/264
/ 7.5 (0.45) |
14/255
/ 9 / (0.61) |
0.75
(0.32, 1.73) |
|
Agnelli
(2003) / 3 month / PE |
No
VKA given after trial began |
|
|
|
Agnelli
(2003) / 6-12 months / PE |
3/165
/ 8 (0.22) |
12/162
/ 23 / (0.35) |
0.68
(0.19, 2.40) |
|
Agnelli
(2001) / 3 months / DVT |
No
VKA given after trial began |
||
|
Agnelli
(2001) / 1 year / DVT |
5/134
/ 11 (0.34) |
16/129
/ 28 / (0.45) |
0.76
(0.28, 2.09) |
|
Kearon
(1999) / 3 months / 1st idiopathic VTE |
No
VKA given after trial began |
||
|
Kearon
(1999) / 24 months / 1st idiopathic VTE |
Subjects
not taken off of VKA during the trial. |
||
|
Kearon
(2004) / 1 month / 1st VTE transient risk factor |
No
VKA given after trial began |
||
|
Kearon
(2004) / 3 months / 1st VTE transient risk factor |
2/81
/ 5 (0.50) |
1/79
/ 8 / (0.16) |
3.11
(0.28, 34.56) |
|
Levine
/ 4 weeks / P-DVT |
No
VKA given after trial began |
||
|
Levine
/ 3 month / P-DVT |
Timing
of recurrences not available |
||
|
Schulman
(1995) / 6 weeks / 1st VTE |
37/441
/ 3.5 (2.40) |
43/408
/ 20.5 / (0.52) |
4.62
(2.95, 7.21) |
|
Schulman
(1995) / 6 months / 1st VTE |
15/451
/ 8 (0.42) |
28/436
/ 16 / (0.40) |
|