Time-restricted salutary effects of blood flow restoration on venous thrombosis and vein wall injury in mouse and human subjects

Abstract

Background:

Up to 50% of patients with proximal deep venous thrombosis (DVT) will develop the post-thrombotic syndrome (PTS) characterized by limb swelling and discomfort, hyperpigmentation, skin ulcers, and impaired quality-of-life. While catheter-based interventions enabling restoration of blood flow (RBF) have demonstrated little benefit on PTS, the impact on the acuity of the thrombus and mechanisms underlying this finding remain obscure. Here in experimental and clinical studies, we examined whether RBF has a restricted time window for improving DVT resolution.

Methods:

First, experimental stasis DVT was generated in C57/BL6 mice (N=291) by inferior vena cava ligation. To promote RBF, mice underwent mechanical de-ligation with or without intravenous recombinant tissue plasminogen activator (rtPA), administered two days after de-ligation. RBF was assessed over time by ultrasonography and intravital microscopy. Resected thrombosed IVC specimens underwent thrombus and vein wall histological and gene expression assays. Next, in a clinical study, we conducted a post-hoc analysis of the ATTRACT pharmacomechanical catheter-directed thrombolysis (PCDT) trial (NCT00790335) to assess the effects of PCDT on VEINES quality-of-life (VEINES-QoL) and Villalta scores for specific symptom-onset-to-randomization (SOR) timeframes.

Results:

Mice that developed RBF by day 4, but not later, exhibited reduced day 8 thrombus burden parameters and reduced day 8 vein wall fibrosis and inflammation, compared to controls. In mice without RBF, rtPA administered at day 4, but not later, reduced day 8 thrombus burden and vein wall fibrosis. Notably, in mice already exhibiting RBF by day 4, rtPA administration did not further reduce thrombus burden or vein wall fibrosis. In the ATTRACT trial, patients receiving PCDT in an intermediate SOR timeframe of 4-8 days demonstrated maximal benefits in VEINES-QoL and Villalta scores (between group difference=8.41 and 1.68 respectively, p<0.001 vs. patients not receiving PCDT). PCDT did not improve PTS scores for patients having an SOR time of <4 days or >8 days.

Conclusions:

Taken together, these data illustrate that within a restricted therapeutic window, RBF improves DVT resolution, and PCDT may improve clinical outcomes. Further studies are warranted to examine the value of time-restricted RBF strategies to reduce PTS in DVT patients.

Introduction

Deep vein thrombosis (DVT), together with its sequelae of post-thrombotic syndrome (PTS), recurrent DVT, pulmonary embolism and chronic thromboembolic pulmonary hypertension, induce considerable disability and mortality worldwide.^1–3 PTS arises in the milieu of persistent thrombus obstruction, vein wall damage and fibrosis, and valvular reflux, ultimately leading to venous hypertension, characterized by limb swelling and discomfort, hyperpigmentation, skin ulcers, and impaired quality-of-life.^2–5 Although therapeutic anticoagulation limits thrombus extension and recurrence, up to one-half of anticoagulated proximal DVT patients still develop symptomatic PTS.^1–3

The “open-vein hypothesis” postulates that rapid removal of luminal thrombus will restore blood flow and improve venous patency, limit the extent of damage to the vein wall and vein valve function, and ultimately reduce PTS.^6–8 Notably, to test this hypothesis, the CaVenT, ATTRACT, and CAVA randomized trials demonstrated limited clinical benefit in reducing PTS outcomes following catheter-based thrombus removal to restore vein patency.^9–12 However, the mechanisms underlying these findings remains unclear. While experimental data demonstrate that aging DVT exhibit lower responsiveness to fibrinolysis^13–15 and that DVT-induced vein wall fibrosis increases over time^16–18, the above clinical trials enrolled patients with a wide range of DVT symptom duration, of up to 14-21 days. To date such studies have not fully analyzed the time-dependent effects of catheter-based therapies on DVT resolution and developing PTS.

Therefore, we hypothesized that earlier restoration of blood flow (RBF) would reduce DVT burden, as well as vein wall inflammation and fibrosis, and would improve clinical PTS outcomes. Experimentally, we generated murine DVT by complete ligation of the inferior vena cava, followed by mechanical de-ligation to promote RBF through DVT, aiming to achieve RBF as with catheter-based therapies. Subgroups of mice further received recombinant tissue plasminogen activator (rtPA) two days after de-ligation, followed by analysis of rtPA effects on occlusive versus non-occlusive DVT. To understand the clinical relevance of these experimental results, we performed an exploratory post-hoc analysis of the ATTRACT pharmacomechanical catheter-directed thrombolysis (PCDT) trial. Specifically, we assessed the effects of restricted timeframes of PCDT on quality-of-life and PTS severity.

Methods

Experimental data, materials and methods will be made available to other researchers on request for the purposes of reproduction or replication from the corresponding author. Experimental studies were approved by the Institutional Animal Care and Use Committee (IACUC) at Massachusetts General Hospital following the National Institutes of Health Guide for Care and Use of Laboratory Animals. For the post-hoc analysis of ATTRACT trial, requests to access the raw dataset for clinical subjects should be addressed to ATTRACT Trial Investigators (http://ClinicalTrials.gov Identifier NCT00790335). The ATTRACT trial was approved by the Institutional Review Boards at all participating centers and all patients provided informed consent. A full description of the materials and methods are presented in the online Supplemental Materials.

Statistical Analysis of Experimental Study

Continuous outcome data are presented as mean (SD) and categorial outcome data are present as proportions. Statistical analyses were completed using GraphPad Prism v8.0 (San Diego, CA). For continuous outcome variables, the Shapiro-Wilk normality test was used to test the distribution of the data. Statistical analysis between two-group comparisons of normally distributed data were applied using the unpaired Students’ t-test, while the Mann-Whitney test was used for not normally distributed data. Statistical analysis among groups with one experimental factor was performed using one-way ANOVA test followed by Tukey’s post hoc test for normally distributed data. For not normally distributed data, the Kruskal-Wallis test followed by Dunn’s post hoc test was applied. To evaluate differences among groups with two experimental factors, data were analyzed using two-way ANOVA followed by Tukey’s test. Correlative trends between the group average of VWCT and group average of thrombus burden was performed using Spearman’s rank-order method.¹⁹ Multiple linear regression analysis was used to identify the relationship between VWCT and IVC wall cells, adjusted for groups. The comparisons of RBF incidence at each time-point were analyzed by Fisher’s exact test. A two-sided P value of 0.05 or lower indicates statistical significance.

Exploratory analysis of the time-dependent effects of PCDT in the ATTRACT trial

We analyzed all 691 acute DVT patients enrolled in the ATTRACT trial (NCT00790335) and further categorized them into three non-prespecified subgroups according to their symptom-onset-to-randomization (SOR) time in days (early, intermediate and late SOR groups, Figure IV in the Data Supplement).¹¹ Data are presented as mean (95% CI). Statistical analyses were completed using R software v3.6 (R Foundation for Statistical Computing) and graphics were created in GraphPad Prism v8.0 (San Diego, CA). The Venous Insufficiency Epidemiological and Economic Study Quality-of-Life (VEINES-QoL) was examined as it is a rigorous patient-reported outcome measure to evaluate QoL, with lower scores indicating a poorer QoL.^11,20,21 The Villalta PTS score provides as it is a validated instrument for diagnosing PTS and quantifying the severity of PTS with a total score range from 0-33 for each leg, with higher scores indicating more severe PTS.^11,22 The modified Marder score (range 0-24) assessed the clot burden of proximal DVT on the basis of quantitative venography, with 0 indicating no DVT and 24 indicating completely occlusive DVT.¹¹ In the ATTRACT trial, the VEINES-QoL and Villalta scores were measured serially at baseline, 10 days, and then at 1, 6, 12, 18, and 24 months. In this analysis, the mean VEINES-QoL and Villalta scores over time or at each follow-up visit were estimated by piecewise linear-regression growth-curve models, with adjustment for strata including the location of DVT (iliofemoral vs. femoral-popliteal DVT) and clinical center and for baseline covariates (age, sex, BMI) by intention-to-treat analysis.^11,21,23,24 To optimize the cut-points for the three SOR subgroups, the two-sided P value was adjusted for multiple testing by the Bonferroni correction (α=0.05/9), such that a P≤0.0056 was considered to indicate statistical significance for the between-group comparisons. To further assess the potential impact of the SOR timeframe on PCDT benefit, significance testing of the interaction between the three SOR timeframes (early, intermediate and late) and treatment (PCDT vs. Control) was conducted by longitudinal mixed effects ANOVA, for the subgroups showing significant between-group differences. A two-sided P value ≤0.05 indicates statistical significance for the interaction analysis.

Results

Development of a new murine occlusive DVT model with time-dependent restoration of blood flow

To date no mouse protocols have reliably generated RBF in a time-dependent manner following DVT formation. Here, we established a new DVT model that can reliably generate RBF through DVT at various time points (Figure 1). Mice with occlusive IVC DVT did not show RBF through the thrombus, and occasionally showed compensatory adjacent collateral veins, confirmed by intravital microscopy (IVM)-based venography (reference standard, Figure 1C). In contrast, nonocclusive murine DVT showed FITC dextran-based blood signal within the IVC thrombus-wall interface. In agreement with IVM venography, doppler ultrasound (DUS) visualized blood flow in the normal and partially thrombosed IVC, as well as flanking collateral veins (Figure 1C). Compared to the reference standard (venography), US showed good sensitivity and specificity of 86.0% and 96.2%, respectively, for detecting IVC blood flow in this model (Figure 1D–1E). In addition, ultrasound showed good agreement with intravital venography (κ=0.72).

Figure 1. A murine IVC DVT ligation/de-ligation model that enables time-dependent restoration of blood flow (RBF) of occlusive DVT.

(A) Experimental flowchart demonstrating surgical IVC knot de-ligation performed at day 2 or day 4, and assessment of RBF using ultrasound (US) or intravital microscopy (IVM). (B) Schematic illustration of initial complete ligation of the IVC on day 0 using a spacer. At day 2 or day 4, IVC knot de-ligation and spacer removal was performed to spur RBF. (C) In vivo assessment of IVC DVT anatomy and blood flow at day 8 in: control mice without DVT; mice with occlusive DVT; and mice with DVT and RBF. Representative IVC images in each group as shown by IVM venography and US. In the RBF group (right images), FITC-dextran signal is seen in between the thrombus and vein wall on IVM (yellow arrows) and on US (yellow arrows). Doppler and pulse wave US reveals opposing blood flow direction in the abdominal aorta (Ao) and IVC, as expected. Absent IVC blood flow on US indicated occlusive DVT, without RBF. Flanking IVC “bridging” collaterals (Co) were also noted in some mice with occlusive DVT. The red solid line demarcates the thrombus. The yellow, red and blue dotted lines/circles indicate the IVC, aorta and collateral vein, respectively. The yellow arrows indicate IVC blood flow. (D) Contingency table comparing ultrasound and venography (reference standard) based on the IVC RBF status. (E) Diagnostic performance of ultrasound compared to venography. Ultrasound showed high sensitivity (86.0%), specificity (96.2%), and positive predictive value (99.0%) for assessing RBF in IVC DVT, as compared to IVM venography. Ultrasound also showed good agreement with venography (κ=0.72). Abbreviations: IVC, inferior vena cava; DVT, deep vein thrombosis; US, ultrasound; IVM, intravital microscopy; H, head; T, tail; Co, collateral vein; R, right; L, left; Ao, abdominal aorta; PV: predictive value.

Mechanical de-ligation restores blood flow through DVT in a time-dependent manner

This study defined early RBF as RBF detected by day 4; and late RBF was defined as RBF detected after day 4. In mice undergoing IVC de-ligation compared to sham de-ligation, the RBF rate increased faster and more completely (~90% RBF rate at day 8) in both day 2 and day 4 de-ligation groups (Figure 2). The day 2 de-ligation group showed significantly faster reperfusion rates at day 4, day 6 and day 8, compared to sham de-ligation group (Fisher’s exact test, ***p<0.001, Figure 2B). Interestingly, in the sham de-ligation group, some mice also demonstrated reperfusion but at slower rates of ~10% RBF rate at day 4 and a ~50% RBF rate at day 8. RBF in sham de-ligation mice occurred through new small veins that formed around the IVC suture knot, as evident by ultrasound and intravital microscopy. Similarly, the day 4 de-ligation group showed faster reperfusion rates at day 8 compared to the day 4 sham de-ligation group (**p<0.01, Figure 2H). The results indicate that de-ligation facilitates RBF in a time-dependent fashion and restores blood flow more effectively than sham de-ligation.

Figure 2. Temporal assessment of mechanical de-ligation, RBF, and venous thrombus burden at day 8.

(A) Experimental flowchart for mice undergoing day 2 IVC DVT de-ligation, followed by RBF assessment using serial US and then venography prior to sacrifice at day 8. (B) The percentage of mice establishing RBF (%) increased steadily after day 2 de-ligation. RBF that developed by day 4 was defined as early RBF, and RBF that developed after day 4 was defined as late RBF. The day 2 de-ligation group demonstrated higher RBF rates starting at 2 days post de-ligation, compared to sham de-ligation; N.S., p>0.05; ***p<0.001; Fisher’s exact test for the comparison of RBF at each timepoint. (C-F) Thrombus burden measures in mice undergoing day 2 de-ligation, as a function of RBF status confirmed by US. N.S., p>0.05; *p<0.05; **p<0.01; one-way ANOVA and Tukey’s test for normally distributed data; Kruskal-Wallis and Dunn’s test for not normally distributed data; n=3-11 animals per group. (G) Experimental flowchart for mice undergoing day 4 IVC DVT de-ligation. (H) The RBF% through DVT increased gradually over time after day 4 de-ligation. The day 4 de-ligation group demonstrated higher RBF rates starting at 4 days post de-ligation, compared to sham de-ligation; N.S., p>0.05; *p=0.052; **p<0.01; Fisher’s exact test. (I-L) Thrombus burden measures in mice undergoing day 4 de-ligation, as a function of RBF status confirmed by US. N.S., p>0.05; one-way ANOVA and Tukey’s for normally distributed data; Kruskal-Wallis and Dunn’s test for not normally distributed data; n=7-10 animals per group. Abbreviations: IVC, inferior vena cava; DVT, deep venous thrombus; h, hour; US, ultrasound; IVM, intravital microscopy; de-li: de-ligation; D, day; RBF, restored blood flow; No., number.

Early but not late mechanical restoration of blood flow reduces thrombus burden

Following day 2 de-ligation, the early RBF subgroup (RBF by day 4, ~50% of the day 2 de-ligation group, Figure 2B) exhibited ~30% reductions in thrombus mass, weight, and width at day 8 (p<0.05 compared to sham de-ligation or de-ligation that did not generate RBF by day 4, Figure 2C–2F). In contrast, late RBF groups (developing RBF between day 4-8) did not manifest significant reductions in thrombus burden parameters compared to mice without RBF. These data indicate that early RBF, but not late RBF, reduced murine thrombus burden as gauged by thrombus mass, weight, and width. Mice (12/291, 4.1%) that died from surgical procedures and mice (25/279, 9.0%) that had no or minimal thrombus at sacrifice (thrombus length ≤0.30 cm compared to average length ~0.80 cm in mice with typical thrombus, Figure 2E) were excluded from analysis as they did not develop the DVT model (Table I in the Data Supplement).

Early but not delayed pharmacological rtPA infusion in occlusive DVT reduces thrombus burden

To induce exogenous fibrinolysis, rtPA and heparin were administered as IV boluses plus a one-hour infusion via a tail vein catheter, on either day 4 or day 6, two days after de-ligation (Figure 3).^14,15 No bleeding complications were noted in the low-dose rtPA group (10 mg/kg), while 7.0% of the high-dose group (20 mg/kg) developed fatal abdominal bleeding (Table I in the Data Supplement). In DVT not demonstrating RBF at day 4, exogenously administered high-dose rtPA at day 4 non-significantly increased the day 6 RBF rate by 38.3%, from 55.6% to 76.9% (p>0.05), and significantly reduced the day 8 thrombus burden (p<0.01; 20 mg/kg rtPA dose; Figure 3B–3F, green symbols), compared to mice with occlusive DVT that received saline. Notably, in mice achieving early RBF by day 4 (i.e. nonocclusive DVT), rtPA administration did not further reduce day 8 thrombus burden (Figure 3C–3F, blue symbols, p>0.05). In contrast to the day 4 findings, high-dose rtPA administered later at day 6 did not improve RBF rates at day 8 (Figure 3H, p>0.05), nor did it reduce thrombus burden measured at day 8, regardless of day 6 occlusive or nonocclusive DVT status (Figure 3I–3L, p>0.05). In a subgroup of mice sacrificed at day 4, mice with and without RBF demonstrated similar thrombus burden (Figure I in the Data Supplement, p>0.05), indicating that the bulk of murine DVT resolution starts at day 4, and is enhanced by early RBF by day 4. Together these data demonstrate that RBF, occurring spontaneously, or induced by de-ligation, or accelerated by exogenous rtPA administration, can reduce murine thrombus burden, but only if delivered by day 4.

Figure 3. Temporal and dose effects of exogenous rtPA therapy on venous thrombus burden at day 8.

(A) Experimental flowchart for mice undergoing day 2 IVC DVT de-ligation and then day 4 rtPA reperfusion therapy. RBF assessment by US was performed at day 4 prior to rtPA administration, and then at day 6 and day 8, followed by sacrifice at day 8 for thrombus burden. (B) High-dose rtPA non-significantly increased the RBF% from 55.6% to 76.9% at day 6; minimal change in the RBF% was evident in the low-dose rtPA group or control group. N.S., p>0.05; Fisher’s exact test for the comparison of RBF at each timepoint. (C-F) Effects of day 4 rtPA infusion on day 8 thrombus burden, as a function of rtPA dose and RBF status. N.S., p>0.05; *p<0.05; **p<0.01; ***p<0.001; two-way ANOVA and Tukey’s test; n=4–23 animals per group. (G) Experimental flowchart for mice undergoing day 4 IVC DVT de-ligation and then day 6 rtPA therapy. (H) RBF% through DVT after day 6 rtPA. High-dose rtPA did not increase the rate of mice achieving RBF. N.S., p>0.05; Fisher’s exact test. (I-L) Effects of day 6 rtPA infusion on day 8 thrombus burden, as a function of rtPA dose and RBF status. N.S, p>0.05; two-way ANOVA and Tukey’s test; n=5-14 animals per group. Abbreviations: D, day; de-li, de-ligation; sac, sacrifice; RBF, restoration of blood flow; IVC, inferior vena cava; DVT, deep venous thrombosis; US, ultrasound; h, hour; rtPA, recombinant tissue plasminogen activator.

Development of early but not late mechanical RBF ameliorates vein wall fibrosis after DVT

To investigate the temporal effects of RBF on vein wall collagen thickness (VWCT), VWCT was measured by both picrosirius red (PSR) and Carstairs’ stain on histological sections.^16–18 Measurements were obtained at 1.2 mm intervals across a 12 mm length spanning the infrarenal IVC to the aortoiliac bifurcation (Figure 4A). Bland-Altman analysis showed good agreement between PSR and Carstairs’ measures of VWCT (Figure II in the Data Supplement). Mice with early RBF exhibited a significant 33.1% reduction in the day 8 VWCT (19.6±4.3 μm vs. 29.3±4.6 μm in the sham de-ligation group, p<0.001, PSR staining, Figure 4C, blue symbols). In contrast, mice without IVC blood flow at day 4 exhibited similar VWCT as the sham de-ligation group (p>0.05, Figure 4C, black symbols). Further analysis of the VWCT profile demonstrated that mice with early RBF by day 4 developed reduced VWCT, as compared to mice achieving RBF after day 4, or to mice undergoing sham de-ligation (p<0.001, Figure 4D).

Figure 4. Temporal assessment of mechanical de-ligation +/- pharmacological rtPA-induced RBF on vein wall fibrosis at day 8.

(A) Schematic illustration of histological section assessment from the infrarenal IVC to iliac bifurcation every 1.2 mm. (B) Representative images of IVC VWCT (vein wall fibrosis) using Carstairs’ and picrosirius red (PSR) stains. Scale bar, 30 μm. (C) Assessment of IVC VWCT at day 8 as a function of RBF status at day 4. The black dotted line shows the mean VWCT (9.4 mm) of the normal IVC. **p<0.01; ***p<0.001; one-way ANOVA and Tukey’s; n=7–10 animals per group. (D) Differences in IVC VWCT at day 8 as function of IVC location and RBF status were further analyzed by two-way ANOVA, followed by Tukey’s. N.S, p>0.05; ***p<0.001, n=7-10 animals per group. (E) Assessment of IVC VWCT at day 8 as a function of rtPA therapy and RBF status. In mice that were already RBF+ at day 4, rtPA did not further reduce the mean VWCT at day 8. N.S, p>0.05; two-way ANOVA and Tukey’s test; n=4-15 animals per group. (F) Day 8 VWCT as function of IVC location and rtPA dose at day 4, analyzed by two-way ANOVA and Tukey’s test. No rtPA groups re-displayed from Figure 4D. N.S, p>0.05; **p<0.01; ***p<0.001; n=4-15 animals per group. (G-J) Correlative trends emerged between the day 8 group average of VWCT and thrombus burden parameters across the 7 groups, with borderline significant associations with thrombus mass, weight and width (n=4-15 for VWCT, and n=4-23 for thrombus burden; r=0.75-0.77, p=0.051-0.066), but not for thrombus length (n=4-23, r=0.23, p=0.61), as analyzed by Spearman’s rank correlation. Scar bar indicates standard error. (K) Vein wall fibrosis measurements in a subgroup of mice sacrificed at day 4 or day 8, as a function of RBF status. Day 8 data re-displayed from Figure 4E. N.S., p>0.05; ***p<0.001; two-way ANOVA with Tukey’s test; n=4–15 animals per group. Abbreviations: VWCT, vein wall collagen thickness; IVC, inferior vena cava; PSR, picrosirius red staining; de-li, de-ligation; d, day; RBF, restored blood flow; sac, sacrifice; rtPA, recombinant tissue plasminogen activator; h, high; avg., average.

Early pharmacological rtPA reperfusion of occlusive but not nonocclusive DVT reduces vein wall injury

In mice with occlusive thrombi, those that received rtPA at day 4, but not at day 6, exhibited reduced VWCT to levels of mice that achieved early RBF (Figure 4F). Notably, rtPA did not further reduce VWCT in mice with non-occlusive DVT (RBF+) at day 4. Overall, the benefits of early rtPA in reducing thrombus burden and vein wall injury appears restricted to occlusive thrombi (RBF-).

When further assessing factors related to the magnitude of VWCT reduction following rtPA administration, thrombus burden and vein wall histological analyses demonstrated trends towards significant relationships between VWCT (as group averages) and thrombus mass, weight, and width (r=0.75, r=0.75, r=0.77, respectively; p=0.066, p=0.066, p=0.051, respectively, Figure 4G–4J), but not for thrombus length (r=0.23, p=0.61, Figure 4I). These results indicate that early RBF groups with lower thrombus burden may develop less vein wall fibrosis.

Notably, despite the initial similar day 4 VWCT in the RBF+ group, VWCT did not increase at day 8 follow-up (Figure 4K). In contrast, in mice without RBF at day 4, VWCT substantially increased at day 8 follow-up, even if RBF was present by day 8 (Figure 4K). The summary data indicate that restoring DVT early blood flow by day 4 by de-ligation, or by early exogenous rtPA for occlusive DVT, proportionately reduced thrombus burden and VWCT.

Temporal relationships between blood flow restoration and vein wall inflammatory and mediators of collagen accumulation

In mice developing early RBF by day 4, the day 8 vein wall demonstrated significant reductions in mRNA levels that encode mediators of inflammation and of extracellular matrix synthesis and breakdown, including F4/80, FSP1, IL-1β, MMP-2, cathepsin B, uPA, and PAI-1 (Figure 5A). No difference in day 4 vein wall RNA transcript levels were evident between mice with and without RBF (Figure III in the Data Supplement). Histological assessment of mice with early RBF exhibited significant reductions both in vein wall F4/80+ macrophages and FSP1+ fibroblasts at day 8 (p<0.01 vs. mice without early RBF, Figure 5B–5D). The relationship between the day 8 VWCT and F4/80+ macrophages was not significant (p=0.24, Figure 5E), but was significant between VWCT and FSP1+ fibroblasts (p=0.022, Figure 5F). The overall data strengthen the concept that early targeting of vein wall macrophage and fibroblast function could reduce vein wall injury.^16,18

Figure 5. Effects of RBF on inflammatory and collagen synthetic mediators in the vein wall following venous thrombosis.

(A) mRNA transcript levels of genes of interest in the IVC wall detected by qRT-PCR at day 8, as a function of RBF status at day 4. Relative fold change compared to the d4 RBF+ group, set at 1.0 (dotted line). Data are presented as mean ± SEM. *p<0.05; **p<0.01; unpaired Students’ t-test for normally distributed data; Mann-Whitney test for not normally distributed data; n=5–9 animals per group. (B) Representative images of IVC wall F4/80+ macrophages and FSP1+ fibroblasts at day 8. Scale bars, 50 μm. (C-D) Day 8 vein wall F4/80+ macrophage and FSP1+ fibroblast cell numbers per 5 HPFs, stratified by RBF status at day 4. ***p<0.001; **p<0.01; unpaired Students’ t-test for normally distributed data; Mann-Whitney test for not normally distributed data; n=7–8 mice per group. (E) The relationship between the day 8 VWCT and vein wall F4/80+ macrophages was not significant (p=0.24). (F) A significant relationship was present between the day 8 VWCT and vein wall FSP1+ fibroblasts (p=0.022). Relationships analyzed by multiple linear regression with adjustment for groups. Abbreviations: D, day; IVC, inferior vena cava; RBF, restoration of blood flow; FSP1, fibroblast-specific protein 1; TNF, tumor necrosis factor; IL, interleukin; MMP, matrix metalloproteinase; Cat B, cathepsin B; TGF, transforming growth factor; uPA, urokinase-type plasminogen activator; PAI-1, plasminogen activator inhibitor type 1; PSR, Picrosirius red; Cars, Carstairs’ IHC, immunohistochemistry; T, thrombus; W, vein wall; L, lumen; VWCT, vein wall collagen thickness; HPF, high-power field.

Pharmacomechanical catheter-directed thrombolysis and quality-of-life in restricted timeframes of DVT patients

Prompted by these experimental results, we next investigated the time-dependent effect of PCDT on PTS outcomes by performing a non-prespecified post-hoc analysis of the ATTRACT trial.¹¹ ATTRACT trial patients were categorized into three non-prespecified subgroups based on their symptom-onset-to-randomization (SOR) time (early, intermediate or late SOR subgroups, Figure IV in the Data Supplement). For individuals in the intermediate SOR 4-9 day subgroups, PCDT associated with a significantly greater improvement in the averaged VEINES-QoL scores at all follow-up visits (p<0.001 vs. anticoagulation alone, Figure 6). Notably, the maximum benefit of PCDT on the VEINES-QoL score occurred in the intermediate SOR day 4-8 subgroup (between-group difference=8.41, p<0.001). Moreover, we observed a significant interaction between the treatment assignment and SOR timeframe stratified as SOR 0-3, 4-8 and 9+ days (treatment × SOR timeframe interaction, p=0.02, Figure 6). Analysis of the Villalta PTS scores in the intermediate SOR 4-8 days yielded similar findings (between group difference=1.68, p<0.001), although the interaction term was non-significant (treatment × SOR timeframe interaction, p=0.13, Figure V in the Data Supplement). All PCDT groups had similar baseline and residual thrombus burden after PCDT, as assessed by the venographic Marder score (p>0.05, Figure VI in the Data Supplement), suggesting that variable residual thrombus burden did not drive the differences in outcomes in this analysis of the ATTRACT trial.

Figure 6. Analysis of different timepoints and their effects of PCDT efficacy for the VEINES-QoL score in the ATTRACT trial.

ATTRACT patients underwent a non-prespecified subgroup analysis categorized as three subgroups based on their symptom-onset-to-randomization timepoint. The mean VEINES-QoL score was estimated by piecewise linear-regression growth-curve models using all available visit assessments from baseline to 24 months, with adjustment for the location of DVT (iliofemoral vs. femoral-popliteal DVT), clinical center and baseline covariates (age, sex, BMI), using an intention-to-treat analysis. To account for multiple testing, a Bonferroni adjusted p-value ≤0.0056 was considered as statistically significant. The benefit of PCDT therapy on VEINES-QoL score was significantly higher in the intermediate SOR timeframe compared to control group. The maximum benefit was evident in the SOR 4–8 day timeframe (p=0.00038). The interaction between treatments × SOR times (G1 vs. G2 vs. G3) was also significant for intermediate SOR 4–8 days (p=0.020). The interaction between treatments × SOR times from G2 vs. the combined G1+G3 group was even more significant for the intermediate SOR 4–8 days (p=0.0089). Abbreviations: SOR, symptom-onset-to-randomization (in days); No, number; PCDT, pharmacomechanical catheter-directed thrombolysis; VEINES-QoL, Venous Insufficiency Epidemiological and Economic Study-Quality of Life; CI, confidence interval; Tmts, treatments; G, group.

To examine the putative time-dependent benefits of PCDT, we assessed the VEINES-QoL score at each follow-up visit (6, 12, 18 and 24 months, Table 1) across the three temporal subgroups. In this analysis, among patients in the intermediate SOR 4-8 day timeframe, PCDT associated consistently with improved VEINES-QoL scores, compared to anticoagulation alone (p<=0.002 at each study visit) among patients in the intermediate SOR 4-8 days timeframe. In contrast, PCDT did not improve VEINES-QoL scores in either the very early SOR group (<4 days) or the late SOR group (>8 days) (p>0.05, all four study visits). Together, these exploratory data provide evidence that benefits of PCDT may be time-dependent. Specifically, PCDT may offer greater benefit over anticoagulation alone when delivered at an intermediate SOR timeframe, compared to early or late SOR timeframes.

Table 1. VEINES quality-of-life score as a function of the time of randomization to PCDT or anticoagulation alone in the ATTRACT trial.

The mean VEINES-QoL score at each follow-up visit was determined in individuals grouped by non-prespecified SOR timeframes (0-3, 4-8 or 9+ day). Higher VEINES-QoL scores indicate a better quality-of-life. Data are shown as mean (95% CI). The mean VEINES-QoL scores were estimated by piecewise linear-regression growth-curve models using all available visit assessments from baseline to 24 months, with adjustment for the location of DVT (iliofemoral vs. femoral-popliteal DVT), clinical center and baseline covariates (age, sex, BMI), using an intention-to-treat analysis. A Bonferroni-corrected P≤0.0056 was considered to indicate statistical significance for the between-group comparison at each time point. The mean VEINES-QoL scores were significantly higher in the PCDT group within the intermediate SOR 4-8 timeframe at each follow-up visit (p≤0.002). Abbreviations: VEINES-QoL, Venous Insufficiency Epidemiological and Economic Study Quality-of-Life; PCDT, pharmacomechanical catheter-directed thrombolysis; No., number; SOR, symptom-onset-to-randomization, in days.

VEINES-QoL score	PCDT group: PCDT + anticoagulant (N=336)		Control group: Anticoagulant (N=355)		(PCDT – Control) group difference
	No. of Patients	Mean (95% CI)	No. of Patients	Mean (95% CI)	Estimate (95% CI)	P-value
Symptom onset to randomization: all acute DVT patients
At 6 months	290	77.94 (75.31, 80.57)	282	74.87 (72.24, 77.50)	3.07 (−0.28, 6.42)	0.073
At 12 months	270	78.46 (75.99, 80.93)	256	75.54 (73.05, 78.03)	2.92 (−0.22, 6.06)	0.067
At 18 months	245	78.99 (76.44, 81.54)	222	76.22 (73.65, 78.79)	2.77 (−0.46, 6.00)	0.094
At 24 months	250	79.51 (76.71, 82.31)	230	76.89 (74.05, 79.73)	2.62 (−1.05, 6.29)	0.16
Symptom onset to randomization: 0–3 days
At 6 months	69	76.29 (71.59, 80.99)	65	80.11 (75.41, 84.81)	−3.82 (−10.17, 2.53)	0.24
At 12 months	59	76.79 (72.16, 81.42)	64	80.78 (76.17, 85.39)	−3.99 (−10.22, 2.24)	0.21
At 18 months	55	77.29 (72.63, 81.95)	56	81.45 (76.79, 86.11)	−4.15 (−10.44, 2.14)	0.20
At 24 months	60	77.93 (73.11, 82.75)	58	82.11 (77.29, 86.93)	−4.32 (−10.83, 2.19)	0.19
Symptom onset to randomization: 4–8 days
At 6 months	127	79.95 (76.40, 83.50)	127	71.62 (68.07, 75.17)	8.33 (3.53, 13.13)	0.001
At 12 months	125	80.45 (77.00, 83.90)	115	72.29 (68.84, 75.74)	8.16 (3.51, 12.81)	0.001
At 18 months	111	80.95 (77.46, 84.44)	99	72.96 (69.45, 76.47)	7.99 (3.27, 12.71)	0.001
At 24 months	111	81.45 (77.77, 85.13)	101	73.63 (69.91, 77.35)	7.83 (2.81, 12.85)	0.002
Symptoms onset to randomization: 9+ days
At 6 months	93	76.82 (72.59, 81.05)	90	75.46 (71.21, 79.71)	1.40 (−4.23, 7.03)	0.64
At 12 months	86	77.33 (73.19, 81.47)	77	76.13 (71.96, 80.30)	1.20 (−4.31, 6.71)	0.67
At 18 months	79	77.83 (73.64, 82.02)	67	76.80 (72.59, 81.01)	1.03 (−4.56, 6.62)	0.72
At 24 months	78	78.33 (73.98, 82.68)	71	77.46 (73.05, 81.87)	0.86 (−4.98, 6.70)	0.77

Discussion

This experimental and clinical study of venous thrombosis demonstrates that the benefits of mechanical and pharmacological restoration of blood flow are time-sensitive. In particular, earlier RBF reduces thrombus burden, vein wall fibrosis, and clinical measures of the post-thrombotic syndrome. In mice achieving RBF by day 4, reductions in thrombus burden and vein wall fibrosis evolved proportionately from day 4 to day 8. Reductions in thrombus burden and vein wall fibrosis paralleled reductions in inflammatory mediators and regulators of extracellular matrix accumulation, and decreased vein wall macrophage and fibroblast content. Furthermore, a new exploratory analysis of the ATTRACT trial showed that for patients randomized between 4 to 8 days after developing symptoms, pharmacomechanical catheter-directed thrombolysis improved VEINES-QoL scores, compared to anticoagulation alone. The overall findings provide experimental and clinical evidence that time-restricted application of reperfusion strategies may improve DVT and PTS outcomes, and thus offers a time-based refinement of the open-vein hypothesis.^1,6,7

PTS frequently causes morbid complications following DVT. Strategies to reduce PTS have embraced the open-vein hypothesis that timely relief of thrombus obstruction will restore blood flow, attenuate vein wall injury and diminish valvular reflux, and reduce venous hypertension and PTS.^1,6,7 Yet despite extensive efforts, neither medical nor interventional therapies have consistently reduced the incidence and severity of PTS beyond standard-of-care anticoagulation.^9–12

Compared to systemic fibrinolysis, local catheter-based approaches to direct fibrinolytic agents and/or extract thrombus have emerged as a safer and more effective alternative to restore blood flow in DVT patients. The randomized CaVenT, ATTRACT, and CAVA trials enrolled acute DVT patients presenting with a wide symptom onset duration of up to 14-21 days.^9–12 These trials did not consistently demonstrate compelling clinical benefits of CDT or PCDT. However, a critical unanswered question is whether a restricted time window for RBF may better identify patients that could benefit from reperfusion strategies, given that both fibrinolytic resistance and vein wall injury exhibit time-dependence.^14–16,18 For example, some patients may rapidly lyse their DVT under anticoagulation alone over 4 days and for which CDT or PCDT would be of no benefit, whereas others may be highly symptomatic with no appreciable lysis under anticoagulation alone, and for which CDT or PCDT may be beneficial.

This study tested the hypothesis that subjects with DVT benefit from early RBF, either by mechanical treatment (de-ligation, as catheter-based angioplasty or thrombectomy is not feasible in mice given the small caliber of their vessels), and with or without exogenous rtPA-based fibrinolytic therapy. Our current results demonstrate a time-dependent benefit of RBF. Specifically, mice achieving RBF by day 4 exhibited a significant reduction in thrombus burden reduction and vein wall fibrosis as measured by vein wall collagen thickness at day 8. Differences in DVT resolution between RBF+ and RBF- mice were not present at day 4, but subsequently evolved during a crucial window between day 4 to day 8. Reductions in day 8 thrombus burden in early RBF mice occurring after de-ligation likely occurred due to a reduction in stenosis severity, an increase in endogenous fibrinolysis and potentially a component of non-critical embolization to the pulmonary arteries.^25–27

Additional studies of mice undergoing de-ligation and/or exogenous fibrinolysis further revealed that benefits in thrombus reduction did not occur if RBF was achieved late by day 6 or day 8. Prior studies suggest that aged DVT may resist fibrinolysis due to increased accumulation of cross-linked fibrin and collagen, and the formation of a shielding neo-endothelial layer that limits contact of fibrinolytic agents to fibrin.^14,15,28 Notably, this study showed that rtPA did not further improve DVT resolution in non-occlusive murine DVT (Figure 3C–3F, 4E-4F, blue symbols). If this observation was validated clinically, it could be advisable to assess the absence of blood flow (e.g. by ultrasound) before administering catheter-based fibrinolytic therapy, which is currently not the clinical standard of care. If blood flow were present, deferring rtPA-based therapies could be pursued without impairing DVT resolution, and thus spare patients risk of hemorrhage from fibrinolytic therapy.

Venous thrombosis formation provokes an intense inflammatory reaction and induces bystander vein wall injury and fibrosis, in an orchestrated process that involves leukocyte invasion, MMPs, NETosis and collagen accumulation.^{4–6,17,25,29,30} Early vein wall fibrosis occurs following direct contact with the thrombus, whereas later vein wall fibrosis may be driven by other factors including thrombus neovascularization and chronic inflammation.^31–34 Yet, the impact of the timing of RBF on vein wall fibrosis remains unknown in DVT resolution. The current findings demonstrate that early RBF reduced vein wall fibrosis, as well as inflammatory and fibrotic mediators (Figure 4–5). As the degree of vein wall fibrosis was similar at day 4 in mice with or without RBF, a crucial time period exists between day 4-8 for inhibiting vein wall fibrosis in murine DVT. One possible mechanism unifying the study findings regarding RBF, thrombus burden, and vein wall scarring maybe related to thrombus biomechanics. Mice with early RBF demonstrated reductions in thrombus width (Figure 2F, Figure 4J), and potentially reductions in vein wall collagen thickness, stretch or distension. Therapies thus that can reduce thrombus width or vein wall distension at early time-points may have the potential to lessen inflammation-driven vein wall fibrosis,³⁵ even if residual non-occlusive/non-distending thrombus remains present.

In addition, the current findings demonstrate that establishing early RBF significantly reduced the number of vein wall macrophages and fibroblasts and demonstrated significant relationship between day 8 VWCT and FSP1+ fibroblasts (Figure 5B–5F). From a cellular perspective, macrophages play a vital role in thrombus resolution and vein wall remodeling in association with MMPs and plasminogen.^16,18 Moreover, they may direct other vein wall cellular activities, such as fibroblasts. In parallel, fibroblasts participate in the healing of thrombus-induced vein wall injury, by contributing to collagen synthesis and vein wall fibrosis.^16,18 Our results strengthen the concepts that anti-inflammatory and anti-fibrotic therapy may ameliorate vein wall scarring and PTS,^5,17,36,37 and that restoration of blood flow is also important in improving DVT resolution, specifically in early-stage DVT. Lastly, when comparing the effects of day 4 exogenous fibrinolysis on thrombus burden versus vein wall fibrosis, reductions in vein wall fibrosis were of lower relative magnitude, suggesting that a substantial amount of irreversible vein wall injury occurs by day 4 in mice without RBF.^38,39 The time window during which DVT-induced vein wall injury becomes irreversible will be important to investigate in the future. The overall results indicate that early RBF can reduce vein wall fibrosis, in contrast to delayed or absent RBF, illustrating a time-sensitive framework to reduce vein wall injury, a key driver of venous hypertension and PTS. More studies are needed to illuminate the role and timing of RBF in reducing vein wall fibrosis.

To date the ATTRACT trial is the largest catheter-based PTS trial in acute proximal DVT patients (N=691), and enrolled patients with a broad duration of symptoms of up to 14 days symptom-onset-to randomization (SOR).¹¹ The overall trial results demonstrated that PCDT did not reduce the risk of PTS nor improve long-term quality-of-life beyond anticoagulation alone.¹¹ Therefore, to further assess the clinical relevance of our experimental findings, we performed an exploratory non-prespecified temporal analysis categorized as three subgroups (early, intermediate and late SOR timeframes, Figure IV in the Data Supplement). The three timeframes were chosen informed by our experimental results and clinical experience indicating that (1) earlier SOR groups may experience a greater clinical benefit from PCDT; (2) some subjects presenting with earlier RBF may avoid vein wall and valve damage before irreversible injury without intravascular mechanical procedures; and (3) in some patients, anticoagulation alone could limit thrombus extension and might suffice to restore blood flow by enhancing endogenous fibrinolysis in some early stage DVT^3,17,40, potentially obviating the need for PCDT.

This post-hoc analysis showed that the between-group difference and the interaction between treatment × SOR timeframe were significant for the VEINES-QoL scores across a range of SOR timeframes (Figure 6, Table 1). Moreover, the largest benefit for PCDT on QoL occurred in the intermediate SOR timeframe of 4-8 days (between-group difference=8.41, p<0.001). Yet, these data warrant several additional comments. PCDT did not confer a benefit over anticoagulation in the earliest SOR group (0-3 days). One possible explanation for this finding is that some early SOR patients might be effectively treated with anticoagulation alone, as observed clinically;^{3,17, 40} PCDT might also induce greater vein and valve injury when applied to earlier-stage DVT.^8,41 Of further note, in the anticoagulation alone groups, patients that presented early on (SOR 0-2 or 0-3 days) had numerically better PTS scores compared to patients presenting later (e.g. SOR 9+ days). As discussed above, early stage venous thrombi are more amenable to fibrinolysis,^14–17,40 and therefore conservative therapies alone may suffice to restore blood flow and stabilize DVT, for patients presenting early. Notably, although the Villalta scores showed a significant difference between-group difference in the intermediate SOR timeframe, the interaction was not significant between treatment × SOR timeframes (p≥0.13, Figure V in the Data Supplement). Therefore, the results of this study indicating that time-restricted PCDT may improve PTS outcomes should be considered as hypothesis-generating. Taken together, PCDT may improve PTS clinical outcomes for patients with DVT presenting in an intermediate timeframe of approximately 4-8 days. More precise assessment of thrombus age employing molecular-structural imaging approaches^{14,15,17,29,42} might further inform the optimal timing of DVT therapies, compared to patient history alone.

The experimental and clinical data presented suggest the need for new randomized clinical trials evaluate the impact of PCDT for PTS outcomes based on the symptom onset to PCDT interval (e.g. <10 days), RBF status, and DVT location (iliofemoral vs. femoral-popliteal). The current results suggest a three timeframe approach. In a first group of DVT patients with an early symptom presentation (e.g. SOR day 0-3), anticoagulation alone could be the initial treatment approach, to inhibit DVT extension and to promote endogenous fibrinolysis and spontaneous RBF. In this group, the presence of RBF could be subsequently assessed on SOR day 3-4 using ultrasound. If spontaneous RBF were present, anticoagulation could be continued, without initial referral for PCDT. If blood flow were absent on SOR day 3-4, such patients could be considered for PCDT as soon as possible (Figure 6, Table 1). In a second group of DVT patients presenting at an intermediate timeframe (e.g. SOR day 4-8), if baseline ultrasound did not demonstrate RBF on presentation, such patients could be referred for PCDT as soon as possible; otherwise if RBF were present, anticoagulation alone could be recommended. In a third group of patients presenting at a late timeframe (e.g. SOR 9+ days), the clinical treatment recommendation could be for anticoagulation alone, given the lack of benefit of PCDT observed in this group. As a next step, the findings of this post-hoc analysis (benefits seen with intermediate, but not early or late PCDT) merit evaluation in existing studies and prospective verification in future trials. If confirmed, then such data could inform a personalized clinical treatment pathway for acute DVT patients at risk of PTS.

This study has limitations. First, although IVC stasis-induced murine DVT generates reproducible thrombus burden and severe vein wall fibrosis-based injury, immediate venous occlusion by complete ligation does not faithfully mimic the clinical situation and does not produce features of PTS such as limb edema and venous reflux are absent.^43–46 Second, as rodents exhibit a 10-fold lower responsiveness to human rtPA,⁴⁷ both supplemental plasminogen and a high dose of rtPA relative to humans required to induce fibrinolysis in mice.^14,15 In addition, concomitant de-ligation and rtPA administration were not performed on the same day due to the risk of fatal hemorrhage. Third, we did not additionally evaluate the effect of anticoagulation before fibrinolysis in this study, as prior studies have shown that anticoagulation can reduce thrombus formation and improve net DVT resolution.¹⁷ Fourth, as this post-hoc analysis of the ATTRACT trial data was not prespecified, and involved multiple iterative testing and the non-blinded threshold selection, the potential for false negative and positive findings exists. Fifth, as dynamic blood flow status was not available in ATTRACT trial, future trials should strive to track blood flow parameters (e.g. absence or present, volume, and rate of blood flow). Sixth, newer thrombectomy devices have the potential to improve the net benefit of catheter-based treatment of DVT patients. Seventh, as de-ligation is not an option in clinical settings, validation of the restoration of blood flow hypothesis would require the use of clinical approaches (e.g., anticoagulation, catheter-based therapies). Eighth, given the imprecision of the clinical symptom onset history to precisely assess DVT age, novel imaging approaches to measure thrombus age^29,48 may provide new insight in future clinical trials.

Conclusions

The benefits of mechanical and pharmacological interventions on DVT resolution appear to be time-restricted for both experimental and clinical subjects. Targeted restoration of blood flow within specific time periods reduces venous thrombus burden and vein wall injury in mice, and has the potential to improve quality-of-life and reduce PTS severity in DVT patients. Further prospective investigation is needed to determine the applicability of these findings to the reduction of PTS and its sequelae in humans.

Clinical Perspective.

What is new?

• Early but not later restoration of blood flow (RBF) can improve experimental DVT resolution, evidenced by reduced thrombus burden and reduced vein wall fibrosis in mice.

• Fibrinolytic therapy can further improve experimental DVT resolution, but only for early stage, occlusive DVT without RBF.

• In a new analysis of the ATTRACT clinical study, the greatest quality-of-life benefit for pharmacomechanical catheter-directed thrombolysis (PCDT) occurred in patients with a symptom-onset-to-randomization timeframe of day 4-8, with no benefit before day 4 and decreasing benefit after day 8.

What are the clinical implications?

• Earlier restoration of blood flow by mechanical and/or pharmacological reperfusion may improve PTS outcomes for a subset of acute DVT patients.

• Prospective clinical studies are needed to determine the optimal timeframe for PCDT to reduce the post-thrombotic syndrome, with further considerations based on the symptom onset to PCDT interval, blood flow status, and DVT location.

• Clinical studies may be indicated to determine whether DVT patients exhibiting RBF can be treated conservatively.

Non-standard Abbreviations and Acronyms

ATTRACT trial

Acute Venous Thrombosis: Thrombus Removal with Adjunctive Catheter-Directed Thrombolysis trial

CAVA trial

Catheter Versus Anticoagulation Alone for Acute Primary (Ilio)Femoral Deep Vein Thrombosis trial

CaVenT trial

Catheter-Directed Thrombolysis for Deep Vein Thrombosis trial

DVT

deep vein thrombosis

DUS

doppler ultrasound

FSP1

fibroblast-specific protein-1

IVC

inferior vena cava

IVM

intravital microscopy

PCDT

pharmacomechanical catheter-directed thrombolysis

PTS

post-thrombotic syndrome

RBF

restoration of blood flow

rtPA

recombinant tissue plasminogen activator

SOR

symptom-onset-to-randomization

VWCT

vein wall collagen thickness

VEINES-QoL score

Venous Insufficiency Epidemiological and Economic Study Quality-of-Life score