Uses of different D-dimer levels can reduce the need for venous duplex scanning to rule out deep vein thrombosis in patients with symptomatic pulmonary embolism

Results 

Table I summarizes the baseline characteristics of the study patients. Four of the 85 consecutive patients evaluated were excluded according to the criteria previously described. Thus, 81 patients, 47 women (58%) and 34 men (42%) with a mean age of 63 years (range, 24 to 95), were eligible for this study. DVT was present in 51 (63%), and 33 (65%) had leg symptoms. An operation or trauma was the most common risk factor for DVT, found in 33 (41%), followed by immobilization in 19 (24%), active cancer in 13 (16%), and congestive heart disease in 11 (14%).

Table I. Baseline characteristics of the study patients

	Parameters	Data
	Characteristics
	Number of patients eligible	81
	Number of patients with DVT	51(63.0%)
	Mean age (y)	63.1±15.7
	Male gender, n (%)	34(42.0%)
	Risk factors
	Operation and trauma	33(40.7%)
	Immobilization	19(23.5%)
	Active cancer	13(16.0%)
	Congestive heart disease	11(13.6%)
	Hormone replacement therapy	10(12.3%)
	Renal failure	7(8.6%)
	Previous history of DVT	6(7.4%)
	Stroke	2(2.4%)
	Long travel	2(2.4%)
	Inflammatory bowel disease	1(1.2%)
	Pregnancy	1(1.2%)

At the initial examination, 25 patients (49%) had proximal DVT, and the remaining 26 (51%) had distal DVT (Table II). No significant difference was found in the proportion of DVT (P = .843) in each location. There were 21 patients (41%) with an isolated venous segment DVT, and the remaining 30 (59%) had multisegment DVT. In the isolated venous segment, distal veins had a significantly higher proportion of DVT compared with proximal veins (P < .0001). Conversely, a significantly higher proportion of DVT with multisegment DVT extended from the proximal to distal veins (P < .0001). Fig 1 shows the detailed anatomic distribution of DVT. Clot burden segment was most frequently found in the soleal veins (67%), followed by the peroneal (35%), popliteal (31%), common femoral (28%), and femoral veins (27%). In contrast, thrombi were less frequently found in the anterior (4%) and posterior tibial veins (6%).

Table II. Distribution of deep vein thrombosis

	DVT distribution	Segments, n (%)	P⁎
	Proximal DVT	25(49.0)
	Distal DVT	26(51.0)	.843
	Total	51(100)
	Isolated segment
	Proximal DVT	2(3.9)
	Distal DVT	19(37.3)	<.0001
	Total	21(41.2)
	Multisegment
	Proximal DVT	23(45.1)
	Distal DVT	7(13.7)	<.0001
	Total	30(58.8)

DVT, Deep vein thrombosis.

⁎ Pearson χ2 test.

View Large Image Download to PowerPoint

Fig 1. Detailed anatomic distribution of deep vein thrombosis. EIV, External iliac vein; CFV, common femoral vein; FV, femoral vein; DFV, deep femoral vein; POPV, popliteal vein; ATV, anterior tibial vein; PTV, posterior tibial vein; PV, peroneal vein; GV, gastrocnemius vein; SV, soleal vein.

Table III lists the distribution of the patients with low, moderate, and high risk according to the calculated PTP score. Of 81 patients, 27 (33%) were classified as low, 38 (47%) as moderate, and 16 (20%) as high risk by PTP. DVT was found in nine patients (33%) in the low-risk group. In contrast, DVT was found in 27 moderate-risk (71%) and 15 high-risk patients (94%). Prevalence of DVT increased as the risk for DVT assessed by PTP score progressed. The results of the D-dimer assay showed no statistically significant differences between the PTP risk groups.

Table III. Prevalence of deep vein thrombosis by pretest clinical probability risk classification

	Pretest clinical probability	N (%)	DVT frequency, n (%)	D-dimer (μg/mL)⁎
	High risk	16(19.8)	15(93.8)	11.36±14.20
	Moderate risk	38(46.9)	27(71.1)	8.51±10.77
	Low risk	27(33.3)	9(33.3)	5.64±6.16
	Total	81(100)	51(63.0)

DVT, Deep vein thrombosis. Continuous values are expressed as mean ± standard deviation.

⁎ The analysis of variance was used for for multiple comparison.

Table IV shows the results of the D-dimer assay. In the moderate PTP patients, there was no significant difference in the D-dimer assay value between patients with positive and negative DVT scan results. In the low PTP patients, however, the difference in the value of the D-dimer assay between positive and negative scan patients was statistically significant (9.99 ± 7.33 vs 3.46 ± 4.20, respectively; P = .008).

Table IV. Results of D-dimer assay between patients with positive and negative scans

	D-dimer (μg/mL)	Venous duplex scanning				P⁎
		Positive		Negative
		No.	Value	No.	Value
	High PTP	15	11.76±14.61	1	5.31	NS
	Moderate PTP	27	10.69±11.99	11	3.14±3.35	.051
	Low PTP	9	9.99±7.33	18	3.46±4.20	.008

PTP, Pretest clinical probability. Values are expressed as mean ± standard deviation.

⁎ Wilcoxon’s nonparametric rank sum test.

The discriminating power of the D-dimer cutoff point of 0.5 μg/mL for low and moderate, and high PTP patients, and entire study population, is summarized in Table V. In all groups, D-dimer testing achieved 100% sensitivity and negative predictive value (NPV) in the diagnosis of DVT; however, as the PTP score increased from low to moderate and to high, the specificity decreased from 33% to 18% to 0%. To improve the discrimination power of D-dimer assay, ROC curve analysis was used to determine different cutoff points in the low and moderate PTP groups (Fig 2 and Fig 3). D-dimer cutoff points of 2.0 and 0.7 μg/mL were selected for the low and moderate PTP groups, respectively. With these specific cutoff points, D-dimer testing still provided 100% sensitivity and 100% negative predictive value in the diagnosis of DVT for both groups (Table VI). However, using these specific cutoff points in the low PTP patients increased the specificity from 33% to 67% (P = .046), whereas in the moderate PTP group, the newly determined D-dimer level did not improve the specificity.

Table V. The discriminating power of D-dimer testing of a single D-dimer cutoff point of 0.5 μg/mL by pretest clinical probability category

	Patient group	Sensitivity (%)	Specificity (%)	PPV (%)	NPV (%)
	All patients	51/51(100)	8/30(26.7)	51/73(69.9)	8/8(100)
	High PTP	15/15(100)	0/1(0)	15/16(93.8)	0/0(NA)
	Moderate PTP	27/27(100)	2/11(18.2)	27/36(75.0)	2/2(100)
	Low PTP	9/9(100)	6/18(33.3)	9/21(42.9)	6/6(100)

PPV, Positive predictive value; NPV, negative predictive value; PTP, pretest clinical probability; NA, not applicable.

View Large Image Download to PowerPoint

Fig 2. Receiver operating characteristics curves analysis for low pretest probability (PTP).

View Large Image Download to PowerPoint

Fig 3. Receiver operating characteristics curves analysis for moderate pretest probability (PTP).

Table VI. Sensitivity, specificity, and positive and negative predictive value for different D-dimer cutoff points

	Patient group	Cutoff point (μg/mL)	Sensitivity (%)	Specificity (%)	PPV (%)	NPV (%)
	All patients	2.0/0.7/0.5	51/51(100)	14/30(46.7)	51/68(75.0)	14/14(100)
	High PTP	0.5	15/15(100)	0/1(0)	15/16(93.8)	0/0(NA)
	Moderate PTP	0.7	27/27(100)	2/11(18.2)	27/36(75.0)	2/2(100)
	Low PTP	2.0	9/9(100)	12/18(66.7)	9/16(56.3)	12/12(100)

PPV, Positive predictive value; NPV, negative predictive value; PTP, pretest clinical probability; NA, not applicable.

Finally, with the use of the single D-dimer cutoff point of 0.5 μg/mL, eight of 81 patients with a low or a moderate PTP score and a normal D-dimer assay required no further investigation, and thus VDS could be reduced by 10% (8/81). However, by using the specific D-dimer cutoff points for the various PTP groups, six additional patients could have been excluded compared with recommended cutoff level of 0.5 μg/mL, and thus VDSs could have been reduced by 17% (14/81).

Discussion 

A combination of a PTP score and D-dimer testing has previously been considered validated as a diagnostic strategy for PE or DVT exclusion,4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19 and this strategy is specifically validated for patients who have low PTP. Anderson et al5 showed that the observed NPV of the D-dimer assay was 100% in the low-probability patients, 94.1% in moderate-probability patients, and 86.7% in high-probability patients. They concluded that the D-dimer assay might be a potentially useful adjunctive test to exclude the diagnosis of DVT in patients with PTP scores. Furthermore, Schutgens et al13 showed that DVT could be excluded in patients with a normal D-dimer concentration and a low and moderate PTP score. They also demonstrated that anticoagulation can be safely withheld in patients with suspected DVT when the combination of a low PTP score and a normal D-dimer concentration is present. Finally, Yamaki et al8 found among each PTP group that D-dimer testing provided 100% sensitivity and 100% NPV in the diagnosis of DVT.

Using D-dimer testing and PTP for exclusion of VTE depends largely on the reliability of the D-dimer test. Several methods are commercially available for measuring D-dimer concentrations. Despite laboratory-based and relatively time-consuming examinations, techniques based on EIA or enzyme linked immunosorbent assays (ELISA) are considered to be the gold standard in determining D-dimer, with the highest sensitivity and NPV for DVT.7, 8 The latex agglutination assay had relatively low sensitivity and specificity for DVT in the past7; however, early latex assays have been replaced by newer assays with improved sensitivity and NPV.12, 13

Schutgens et al14 evaluated four new D-dimer assays and a classic ELISA in outpatients with symptoms suggestive of DVT and concluded that the Tina-quant latex assay (Roche, Mannheim, Germany) had the highest NPV of 98%, and a high sensitivity of 99% using the standard cutoff value.14 Recent studies have also shown that a rapid method of D-dimer assay on the basis of whole blood agglutination that could be performed at the bedside is useful in excluding patients without DVT.4, 5, 6, 15, 16, 17, 18 Wells et al15 evaluated 214 patients with clinically symptoms suggestive of DVT using the SimpliRED D-dimer assay (AGEN Biomedical, Brisbane, Australia), which had a sensitivity of 93% for proximal DVT and a negative predictive value of 98%. Di Nisio et al18 studied the combined use of PTP and D-dimer assay and found that D-dimer testing achieved a NPV of 100% and 97% among cancer patients with low PTP or low-moderatePTP scores, whereas corresponding NPV in the absent of malignancy was 98% and 97%, respectively.

Another factor that could affect the discriminating power of the studies for VTE is the cutoff value of the D-dimer test. The sensitivity could be improved by lowering the cutoff value, but the subsequent decrease in specificity would lead to a large number of false-positive results. D-dimer cutoff values with very high specificity provide fewer false-positive results, but they are less sensitive for VTE and cannot be used to exclude the disease in all patients.

Against this background, Linkins et al19 used data from a previously published study of 571 patients and reported that varying the D-dimer cutoff values according to PTP score would exclude VTE in more patients than using the single D-dimer cutoff point.19 In that analysis, they found in the high PTP score group that a D-dimer cutoff of 0.2 μg fibrinogen equivalent units (FEU)/mL was required to achieve a NPV of 98% or higher. Similarly, in the low PTP score group, a D-dimer cutoff point of 2.1 μg FEU/mL achieved a NPV of 98% or higher. With this strategy, the number patients with false-positive results also dropped from 89 to nine in the low PTP group (n = 205).

Patients with PE usually present with an elevated D-dimer assay, and D-dimer values <0.5 μg/mL have an almost 100% NPV for DVT exclusion.21, 22 Moreover, Perrier et al22 found that a D-dimer level >4 μg/mL was associated with a specificity of 93% for PE detection.22 Later, Kucher et al23 studied 85 PE-positive patients and found that the D-dimer level should be >7 μg/mL to achieve the specificity of 100%.23 In the current study, however, the D-dimer value in patients with no DVT was much lower than 7 μg/mL. The possible explanation for this is that only cardiopulmonary stable PE patients were selected for this study. Even patients with small PE were referred to our department. Elevated D-dimer levels are present in many other circumstances, however, including advanced age24 and during postoperative periods.25 Therefore, a decision-making positive diagnosis threshold cannot be defined with D-dimmer alone.12

The presence of proximal DVT is considered to be a part of PE diagnosis.12 In contrast, the screening for DVT in patients with proven or suspected symptomatic PE still remains debated.26, 27 Giachino et al26 reported that 13% of fatal pulmonary emboli originated only from calf DVT and concluded that isolated DVT do produce fatal PE. However, Gottlieb et al28 found that VDS of the calf is not mandatory at the initial evaluation to identify patients at risk of clinically important PE. Even in the proximal veins, Girard et al29 found no prognostic significance for screening in patients with proven PE. These reports, however, studied the risk of relatively early thromboembolic sequelae alone.

To evaluate late post-thrombotic complications that affect the rest of a patient’s quality of life, the initial distribution and extent of DVT may be important.30, 31 Concerning distribution of DVT, our results are comparable with previous studies. Girard et al29 found that the prevalence of DVT in PE patients was 60%, which was similar to our results, but the prevalence of distal DVT in their study was much lower than in ours. We have been trying to rule out even small, isolated calf vein thrombi as well as proximal vein thrombosis because a missed diagnosis directly reflects the low sensitivity of VDS in the diagnosis of DVT, leading to the large volume of negative VDS studies. Our present study showed no significant difference in the proportion of DVT between proximal and distal veins. Furthermore, isolated calf DVT plays an important role in the development of PE as well as in multisegment proximal DVT. Detailed DVT detection according to PTP scores and D-dimer assays may thus be more logical than detection of proximal DVT in all patients with PE. Early detection of DVT is also useful for selecting patients who require much longer follow-up studies.

The patients in this current study already had PE, and this could affect the results of D-dimer testing. When a single cutoff point of 0.5 μg/mL was used, 24% of the patients (9/38) with moderate, and 44% (12/27) with low PTP scores showed false-positive results. When a specific D-dimer cutoff point of 2.0 μg FEU/mL was selected for low PTP patients, the number of patients with false-positive results dropped from 44% to 22% (6/27), a 22% change. A sensitivity and NPV of 100% for excluding DVT in this group was still achieved. Reducing the number of false-positive results could decrease the number of unnecessary VDS, leading to saving both time and money. In contrast, a specific D-dimer cutoff point of 0.7 μg FEU/mL did not improve the discrimination power in patients who had moderate PTP scores compared with a single cutoff level of 0.5 μg/mL.

Our study has a possible limitation. The NPV of the different D-dimer values appeared equal to that of a single D-dimer cutoff point. In addition, although a specific D-dimer level substantially decreased the number of false-positive results in the low PTP group, within the group of patients with moderate PTP, a specific D-dimer level did not improve the specificity, leading to a relatively high percentage of false-positive results among this group of patients. Furthermore, the utility of D-dimer assay was not confirmed in patients with high PTP scores. This study thus suggested that the D-dimer test may have a high NPV and a high sensitivity in PE patients with low PTP scores but has scarce utility within the moderate and high PTP groups.

These results potentially provide a new strategy for the diagnosis of DVT in patients with PE, as demonstrated in Fig 4. Patients who have a low PTP score and D-dimer level of <2.0 need no further investigation, whereas patients with a low PTP score and D-dimer level of >2.0 undergo a single VDS at presentation. Patients who have a moderate PTP score and a normal D-dimer level need no further investigation, and patients who have a moderate PTP score with an elevated D-dimer level should undergo a single VDS at presentation. Finally, patients with high PTP score should undergo VDS immediately, and those with an initial negative scan result should have a repeat VDS after 1 week because these patients have quite high prevalence of DVT.

View Large Image Download to PowerPoint

Fig 4. Diagnostic strategy for the diagnosis of deep vein thrombosis (DVT) in patients with pulmonary embolism (PE).

Conclusion 

A combination of specific D-dimer levels and clinical probability scores is most effective in patients with low PTP scores in excluding DVT. In patients with a moderate PTP score, the recommended cutoff point of 0.5 μg/mL may be preferable; however, in patients with high PTP, D-dimer testing results are not useful in excluding those with lower extremity DVT. Thus, these results show that different D-dimer levels for patients with different PTP scores is more useful than a single cutoff point of 0.5 μg/mL in excluding DVT in patients with confirmed PE.

Author contributions