Distal DVT: worth diagnosing? Yes
S.M. Schellong, Division of Angiology, University Hospital Carl Gustav Carus, Fetscherstraße 74, 01307 Dresden, Germany. Tel.: +49 351 458 3659; fax: +49 351 458 4359; e-mail: firstname.lastname@example.org
Summary. Much of the argument for or against diagnosis of distal deep vain thrombosis (DVT) depends on the extra effort that has to be spent on it. This review presents the data on ultrasound of paired calf veins and calf muscle veins (distal ultrasound) in terms of protocols, feasibility, reliability and expected findings. In summary, provided there is adequate and anatomically sound training of sonographers, distal ultrasound is a valid, 4-minute procedure, which can easily be added to the examination of proximal veins. The second part of the review refers to the pathophysiology of ascending DVT, which is the most common type. Adequate patient care in terms of benefit, harm and cost includes a single non-invasive examination followed by risk adopted treatment allocation. This concept ideally should be valid for any type of DVT. The data extending this concept to distal DVT can only be derived from studies that look closely at this entity (i.e. in fact diagnose distal DVT). Even before these data are available, diagnosing distal DVT at least doubles the number of symptomatic patients in which signs and symptoms can be ascribed to a definitive diagnosis, which in itself is a benefit for patient care.
This review deals with the diagnostic work-up of symptomatic patients [i.e. patients with a clinical suspicion of deep vein thrombosis (DVT) on the basis of signs and symptoms confined to the leg]. The question is whether the gain of information yielded by ultrasound of examining paired calf vein thrombosis and calf muscle vein thrombosis (distal ultrasound) justifies the extra effort of performing it. As a basis for any further discussion, clarification is needed with regard to ultrasound anatomy and ultrasound protocols (see Fig. 1) .
The landmark study by Lensing et al.  from 1989 used a compression ultrasound protocol examining the groin and the popliteal fossa. Restriction to these two points (2-CUS) was based on venographic evidence that proximal DVT in symptomatic patients, if present, will be detectable at one or both of these regions . With growing experience in venous ultrasound, the examination was extended to all segments from the groin to the distal part of the popliteal fossa, thereby covering the entire length of the femoral vein and the so-called trifurcation area (E-CUS). This protocol has been used in many diagnostic studies since the 1990s .
Several authors proposed protocols examining systematically distal veins as well. They differ in which Doppler modalities they use, mainly color Doppler, and whether they aim for longitudinal sections of distal veins [5–8], or rely entirely on compression ultrasound restricted to cross-sections of the respective venous segments . For the latter approach the term complete compression ultrasound (C-CUS) was coined. Regardless of these differences, most authors agree that examination of anterior tibial veins, albeit technically not difficult, is not mandatory. In contrast, all authors favoring distal ultrasound do examine calf muscle veins.
Several investigators have reported feasibility data for distal venous ultrasound. Feasibility relates to two different issues, the first being the time needed to screen distal veins, the second being the rate of inevaluable patients.
The time needed to perform distal ultrasound depends on the protocol. Obviously, utilizing longitudinal sections of veins and Doppler modalities requires more time than restricting the examination to compression ultrasound in cross-sections. Thus, Stevens et al.  reported an additional time of 15 to 30 min for distal ultrasound, in accordance with Gottlieb and Subramaniam, who took about 10 min and 10–15 min, respectively [6,8]. In contrast, Schwarz et al.  measured an average of 4 min extra time per leg to complete the distal part of the examination. The examination was shorter in negative patients and longer in positive patients. Of course, time used also depends on personal skills and training. In the experience of the Dresden University ultrasound laboratory, a beginner in vascular ultrasound needs around 200 supervised C-CUS examinations in order to perform adequately. Provided that it is based on sound understanding of venous anatomy it is not more complicated than for other vascular ultrasound examinations.
All authors studying systematically distal ultrasound consistently reported very low rates of inadequate examinations, ranging from 0.4% to 1.4%. When taken together, the four studies [7,8,11,12] showed an inevaluability rate of 0.98% (95% CI, 0.68–1.37). As all studies reported consecutive patients, they are likely to reflect everyday practise in a referral center interested in distal ultrasound.
Because for venography in symptomatic patients the rates of technical failures only average 1% [2,13], not including inadequate studies, the figures for distal ultrasound do not suggest that inevaluability has to be considered a major problem. It is important to note that obesity in itself is no obstacle for distal ultrasound. The same is true for epifascial edema. However, interstitial muscle edema is a hindrance, as is the inability of the patient to sit upright.
A major concern regarding distal ultrasound is the validity of the test. Data are available for both external and internal validity.
Both components of accuracy as a measure of external validity may be relevant for patient safety. There are two meta-analyses integrating studies comparing ultrasound in symptomatic patients with venography as the gold standard. Kearon et al.  calculated a sensitivity for distal veins of 73% (95% CI, 54–93%), while a specificity of 94% (95% CI, 90–98%) was given for proximal and distal veins together. A more recent meta-analysis  differentiated between ultrasound protocols and revealed that protocols utilizing triplex modes were more sensitive for distal veins at 75% (95% CI, 68–82%) and slightly less specific at 94% (95% CI, 93–96%) compared with compression ultrasound only, with sensitivity and specificity of 57% (95% CI, 49–66%) and 98% (95% CI, 97–98%), respectively. If over-treatment is a concern, the more specific protocol seems to be more appropriate. However, the authors identified additional factors causing heterogeneity between studies. For this reason, it is justified to look also at single dedicated centers with clearly described and well-structured examination protocols. Elias et al.  were the only ones to present early accuracy data as well as later outcome studies, thereby documenting their detailed interest in distal ultrasound. Accordingly, in a large validation cohort they found sensitivity and specificity of 98% and 95%, respectively.
Schwarz et al.  investigated the internal validity of compression ultrasound in terms of inter-observer agreement. With an overall kappa-coefficient of 0.94 for any DVT they reported separate figures for different venous segments ranging from 1.0 for proximal veins to 0.9 for distal and 0.74 for muscle vein thrombosis, which fits well with earlier data by Barrellier et al. . Obviously, to yield such figures focused training is required.
All studies investigating the role of distal ultrasound have reported the extent of clots detected in their patients. Throughout the differing protocols the rates range from 40% to 70% for proximal and from 30% to 60% for distal DVT, with an average of 50% for each. Only one study reported the rate of muscle vein thrombosis separately, which accounted for 50% of all distal DVTs (Table 1). The overall frequency of DVT in the different cohorts is mainly attributable to the referral pattern to the respective center, rather than to the sensitivity of the ultrasound protocol.
Table 1. Proportion of positive findings in cohorts of symptomatic patients undergoing ultrasound of proximal and distal veins
For venography studies lower proportions of distal DVT have been reported . Two reasons may be given. First, routine venography is less likely to detect calf muscle vein thrombosis as this requires a special examination protocol. Secondly, we may speculate that referral patterns have changed over time. As awareness of DVT has increased over the past 10 years (i.e. the time that venography studies were used) and as sonography is a non-invasive technique, patients are sent more liberally and possibly earlier in the course of the disease for objective testing.
Nevertheless, in the framework of current referral attitudes, distal ultrasound roughly doubles the number of patients in whom signs and symptoms can be attributed to a detailed DVT diagnosis. Even before treatment options are considered, this has to be interpreted as a benefit for patient care. It offers the opportunity for detailed counseling and informed decision-making. In consequence, it avoids additional diagnostic tests searching for alternative diagnoses. This is even more true if the number of not DVT-related diagnoses is taken into account, which could easily have been detected by calf ultrasound (i.e. muscle fibre rupture or localized hematoma of the calf). The frequency of these has been estimated to be around 10% .
For proximal DVT, it is firmly established that there are two reasons for treatment: prevention of clinical PE, and prevention of post-thrombotic syndrome (PTS) . Both reasons have been questioned for distal DVT. Regarding PTS, incomplete data and clinical experience strongly suggest that the risk is considerably lower than in proximal DVT, if there is any . Regarding pulmonary embolism (PE), two different views prevail: (i) The thrombus mass in calf veins is well fixed and too small to produce clinical PE. This is the basis for the view that distal DVT does not cause clinical PE. (ii) Distal DVT is able to propagate to proximal DVT by thrombus apposition. If thrombus growth is extensive and fast enough the most recently formed clot will cause clinical PE when breaking off. This is the basis for the view that distal DVT may cause clinical PE.
In fact, about 90% of DVT are of the ascending type . The potential for embolism depends on the speed and the extent of the dynamic, ascending clot growing process. Sensu strictu, almost all clinical PE originate from, formerly, distal DVT. Only the remaining 10% are derived from clots without connection to the lower leg veins (e.g. isolated iliac vein thrombosis, transfascial great or small saphenous vein thrombosis, subclavian vein thrombosis, or catheter-related thrombosis). Thus, there is good reason to stop thrombus growth by anticoagulation once distal DVT has been detected.
The universal application of this concept is challenged by the observation that there is a high percentage of DVTs that do not propagate constantly but lose their dynamics and become ‘abortive’ at some time. This is true for proximal DVT, but it seems to be more relevant for distal DVT, and even more for isolated calf muscle vein thrombosis. There are only a small number of natural, history studies that allow estimation of the ratio between propagating vs. abortive distal DVT. A compilation of these studies has recently been presented by Righini et al. . Despite largely differing methodologies between studies the rates of DVT propagating from distal to proximal fell into the range of 0% to 25%, accounting for a mean of about 10%. An earlier analysis had made an estimate of about 20% . That means that in 80% to 90% of patients with distal DVT no anticoagulation is needed. The question arises how to identify the remaining 10% to 20%? There are two approaches.
(i) Do not look for distal DVT but wait for those propagating to proximal. Depending on the incidence of proximal DVT in the referral cohort this means 65% to 85% second look ultrasounds. With a positive yield between 0.7% and 1.3% [4,15] this approach is clearly not cost-effective . In addition, patient follow-up is incomplete, ranging from a favorable 96% in prospective trials  down to 30% in a real world setting . Finally, the potential hazard of this approach has been documented in studies where clinically relevant and even fatal PE occurred before the second look ultrasound .
(ii) The alternative approach is to look for distal DVT at the first presentation and then to stratify patients for their risk of clot propagation. Treatment would be offered only to those at risk. Treatment intensity and duration could be tailored according to the risk profile. The clinical risk profile for clot propagation seems very likely to be the same as for clot formation. For muscle vein thrombosis (MVT), a pair of studies illustrates this concept. In a cohort study, Schwarz et al.  found a rate of progression into the deep venous system of 25% with compression therapy only; a 10-day course of low-molecular weight heparin (LMWH) reduced this rate to 0%. Notably, in this study there was a high proportion of patients with ongoing risk factors such as active cancer and immobilization. When following a prospective randomized design  the authors recorded propagation rates of 2% without any difference between LMWH and untreated control. However, the proportion of patients with ongoing risk factors was remarkably low.
To follow a risk assessment and risk-adopted therapy approach needs more data regarding progression rates in different patient populations. Such data need well-characterized patient cohorts, and finally a randomized, controlled trial. However, for this purpose there has to be a broad consensus about the anatomic definitions of distal DVT, including MVT, and how to diagnose it reliably. As has been pointed out in the above sections this requires trained sonographers for whom distal ultrasound examination is part of their daily practise.
Until we have such data, everyone who is really interested in the natural history of DVT should opt for one of the justified treatment strategies (no treatment, full treatment for 3 months, in-between regimens), adhere to it for a period of time long enough to form a cohort, and then report on that cohort. This will broaden the spectrum of options from which the study arms for a randomized controlled trial can be chosen.
Of course, all this requires diagnosis of distal DVT.
Thanks to S. Hochauf for assistance with the preparation of the manuscript.
Disclosure of Conflict of Interests
The authors state that they have no conflict of interests.