1040

Venous ulceration is the final outcome of lowerextremity chronic venous insufficiency in most cases.It is a debilitating recurrent complication with anestimated prevalence rate of 0.1% to 0.2% of the

population in developed countries.1 The pathophys-iologic mechanisms at the microcirculatory levels arestill not well understood.

In previous experiments in our laboratory, wedemonstrated that the fibroblasts that were culturedfrom venous ulcers exhibited reduced growth ratesand were morphologically different (larger in size)when compared with the normal fibroblasts thatwere isolated from the ipsilateral limb of the samepatient.2 We further demonstrated that the woundfibroblasts in vitro displayed more characteristics ofsenescence, such as increased senescence associatedβ-galactosidase activity (SA-β–Gal) and an increasedlevel of fibronectin production,3 which are both

Fibroblasts cultured from distal lowerextremities in patients with venous refluxdisplay cellular characteristics ofsenescenceManuel V. Mendez, MD, Andrew Stanley, MD, Tania Phillips, MD, MelissaMurphy, BS, James O. Menzoian, MD, and Hee-Young Park, PhD, Boston, Mass

Purpose: Venous reflux precedes the development of venous ulcers. Our earlier workshowed that the fibroblasts that are cultured from these wounds display more charac-teristics of senescence. We evaluated fibroblast senescence in patients with venous refluxbut without ulcers to further investigate the role of venous reflux in the predispositionto venous ulcers.Methods: Fibroblasts that were isolated from skin biopsy specimens of the “gaiter” area(distal) and of the ipsilateral thigh of the same patient (proximal) were compared.Twelve patients with venous reflux (9 patients in clinical, etiologic, anatomic, and patho-logic classification 4; 3 patients in classification 5) with an average venous filling indexof 5.45 mL/s and 4 patients without venous reflux were enrolled in the study. Thegrowth rates, the response to basic fibroblast growth factor (b-FGF), and the senescencemarkers (β-galactosidase activity at a pH level of 6, unstimulated fibroblasts fibronectinprotein, and messenger RNA levels) were determined for each cell population.Results: The number of senescence-associated β-galactosidase positive cells (8.3% ± 1.9%vs 2.2% ± 0.8%; P = .008) and the level of cellular fibronectin protein (455.7 ± 80 vs 210± 51; P = .04) and messenger RNA (16.8 ± 6.8 vs 13.5 ± 5.7; P = .042) were significantlyhigher in the distal fibroblasts as compared with the proximal fibroblast cultures. Thegrowth rates of the distal fibroblasts were lower when compared with the proximalfibroblasts (15,746 ± 4287 cells/day vs 29,550 ± 5035 cells/day; P < .002) but were notdifferent in the presence of b-FGF (41,717 ± 9542 cells/day vs 47,030 ± 6133 cells/day;P = .53). In the patients without venous reflux, no site differences were noted in thegrowth rates or the senescence markers between the proximal and distal fibroblasts.Conclusion: Distal fibroblasts that are isolated from patients with venous reflux displaymore senescence characteristics than do proximal fibroblasts and have significantly lowergrowth rates. Despite senescence, b-FGF restored the distal-fibroblasts growth rate tothat of the stimulated proximal fibroblasts, which proposes a therapeutic role for b-FGF.These changes precede ulcer formation and suggest a mechanism that is focal and intrin-sically related to venous reflux. (J Vasc Surg 1998;28:1040-50.)

From the Section of Vascular Surgery, Boston University MedicalCenter.

Presented at the Twelfth Annual Meeting of the Eastern VascularSociety, Newport, RI, May 1–3, 1998.

Reprint requests: James O. Menzoian, MD, Department ofSurgery, Boston Medical Center, 88 E Newton St, D-506,Boston, MA 02118.

Copyright © 1998 by The Society for Vascular Surgery andInternational Society for Cardiovascular Surgery, NorthAmerican Chapter.

0741-5214/98/$5.00 + 0 24/6/93327

well-described characteristics of the human senes-cent fibroblast.4,5

The accumulation of the senescent fibroblastswithin the ulcer environment could affect the heal-ing process through several mechanisms, such as theoverexpression of proteolytic activity (an increase inthe production of matrix metalloproteinases and theunderexpression of protease inhibitors), dermisimmunomodulation (the overexpression of inter-leukin-1,7 and the underexpression of interleukin-6and interleukin-8 after TNF-α challenge)8, and analtered matrix deposition (the increased productionof abnormal fibronectin)9.

It is not known whether the fibroblast abnor-malities that are found in venous ulcers are a preex-isting local phenomena induced by intrinsic mecha-nisms, such as chronic venous hypertension, orwhether these changes are induced by the extrinsicmechanisms associated with the ulcer environment,such as injury, bacterial colonization, or infection.

We hypothesized that chronic venous hyperten-sion through intrinsic mechanisms, such as thesequestration and activation of macrophages or mastcells within the dermal compartment, affect thefibroblast function and lead to a relative accumula-tion of senescent fibroblasts before ulcer formation.

This study was designed to examine whetherfibroblasts that were isolated from the dermis of thedistal lower extremities of patients with venousreflux but no ulcers display abnormalities similar tothose that are found in the ulcer fibroblasts.

METHODSPatients. The patients with symptomatic venous

disease were interviewed and examined. The evalua-tion of the venous disease was performed clinically bya trained vascular surgeon with the guidelines thatwere prepared by the Executive Committee of theAmerican Venous Forum (Clinical Classification andGrading of Chronic Venous Disease),10 and venousreflux was determined in the noninvasive vascularlaboratory by means of air plethysmography andvenous duplex scan examinations. The patients inwhom concomitant arterial disease was suspectedbecause of absent pulses were excluded from thestudy. The patients who were taking immunosuppre-sive medication also were excluded from the study.All the patients with diabetes and the patients withknown HIV-positive status were excluded as well.The Institutional Review Board at Boston MedicalCenter approved this study, and informed consentwas obtained for all patients who were enrolled in thestudy. In addition, the healthy volunteers without

JOURNAL OF VASCULAR SURGERYVolume 28, Number 6 Mendez et al 1041

venous disease were interviewed and examined andunderwent noninvasive vascular examination forassessment of asymptomatic venous reflux.

Assessment of venous reflux. All the patientsunderwent noninvasive vascular testing at the BostonMedical Center Vascular Laboratory. A venous fillingindex, as determined by means of air plethysmogra-phy, that was greater than 2 mL/s was used as thecriteria for the diagnosis of venous reflux. Venousduplex scan evidence of significant venous retrogradeflow or of anatomical valvular incompetence of thesuperficial, deep, or perforating venous system alsowas considered as inclusion criteria.

Materials. Dulbecco’s modified Eagle’s medium,trypsin, and Trizol reagents were purchased fromGibcoBRL Life Technologies, Inc (Gaithersburg, Md).The monoclonal anticellular fibronectin antibody waspurchased from Sigma (St Louis, Mo). The recombi-nant human basic fibroblast growth factor (b-FGF) was obtained from Collaborative Research. The ECL Western blotting detection system kit was pur-chased from Amersham Life Science (ArlingtonHeights, Ill). The bovine calf serum was obtained fromHyclone Laboratories, Inc (Logan, Utah). The nitro-cellulose and nylon membranes were purchased fromAmersham (Arlington Heights, Ill), and the oligolabel-ing kit was from Pharmacia LKB Biotechnology(Piscataway, NJ). The radiolabeled nucleotide, d-CTP,was from New England Nuclear (Boston, Mass), andthe X-Gal (5-bromo-4-chloro-3-indolyl β-D-galacto-side) was purchased from Sigma.

Fibroblast isolation and culture. From eachpatient or volunteer, a biopsy was obtained from theuninvolved skin of the proximal thigh and from theuninvolved skin of the “gaiter” area of the ipsilaterallimb. The techniques that were used to isolate and cul-ture the dermal fibroblast are described in detail else-where.2 In brief, the biopsies were treated with a 5-minute povidone-iodine bath that was followed by a 5-minute bath in a 70% ethanol solution. An overnighttrypsin incubation then was performed at 4°C (trypsinsolution, 1 µg/mL), which allowed for the separationof the dermal tissue from the epidermal and adiposetissue. The dermal tissue was isolated with sterile con-ditions, cut sharply into 1-mm to 2-mm fragments,and placed in etched plastic tissue culture dishes(Fisher, Pittsburgh, Pa). A complete medium that con-sisted of Dulbecco’s modified Eagle’s medium with10% calf serum and antibiotic solution (penicillin,methicillin, fungizone, and streptomycin) was addedto support cell growth. Two separate fibroblast popu-lations were obtained from each patient or subject—the proximal fibroblasts and the distal fibroblasts. The

Fibronectin protein analysis (immunoblot).An immunoblot analysis (Western blot) for cellularfibronectin was performed after the extraction of thematrix proteins from the fibroblast culture dishes.The second passage fibroblasts were allowed toachieve an 89% to 90% confluence in 60-mm culturedishes. The urea-based extraction of the matrix pro-teins then was performed by lysing the cells into a 4-mol/L urea buffer and homogenate set for 1 to 2hours at 4°C. The total amount of protein was quan-tified with the Bradford protein assay technique,which is described elsewhere. Ten to 30 µg of celllysate protein was separated with a 7.5% sodiumdodecyl sulfate-polyacrylamide gel electrophoresis.The same amount of total urea extracted protein wasloaded for both the proximal fibroblasts and the dis-tal fibroblasts cultures from each patient. The pro-teins then were transferred from the gel to a nitro-cellulose membrane. The membrane was preincu-bated in 100% Blotto (5 g of nonfat dry milk in 100mL of PBS) with shaking for 1 hour at room tem-perature. This was followed by the incubation with amouse antihuman monoclonal antibody against cel-lular fibronectin (0.2 to 1 µg/mL in 10% Blotto) at4°C. At the end of the incubation, the membranewas washed extensively with PBS, which contained0.5% Tween-20, and then was reacted with the sec-ondary goat antimouse antibody labeled with horse-radish peroxidase. Chemiluminescent reaction withECL Western blotting analysis system was used todetect the fibronectin bands with autoradiographyexposing membrane to Kodak X-Omat film.

Fibronectin mRNA detection. The total RNAwas extracted from the passage 3 fibroblasts at an80% to 90% confluence in 60-mm culture dishes bylysing the cells with the Trizol reagent (1 mL per 60-mm dish). The lysate was equilibrated at room tem-perature for 5 minutes and was extracted with chlo-roform (0.2 mL per 1 mL of Trizol reagent). TheRNA was precipitated by adding 0.5 mL of iso-propyl alcohol per 1 mL of Trizol reagent. Equalamounts (7 to 10 µg) of total cellular RNA from theproximal fibroblasts and the distal fibroblasts of thesame patient were separated electrophoreticallythrough a 1% agarose gel that contained 2.2 mol/Lformaldehyde, were blotted on a Hybond-N nylonmembrane, and were immobilized with short ultra-violet light illumination. The blot was prehybridizedfor 24 hours and incubated with the specific radiola-beled fibronectin insert (2.0 × 106 to 3.0 × 106

cpm/mL hybridization solution) for 24 hours. Themembrane were washed twice for 30 minutes with2× standard saline citrate solution at 65°C and then

JOURNAL OF VASCULAR SURGERY1042 Mendez et al December 1998

cells then were passed into 2 separate tissue culturedishes (passage 1). Once an 80% to 90% confluencewas reached, one plate was used to freeze the fibrob-lasts with dimethyl sulfoxide solution, and then thecells were stored at –140°C in liquid nitrogen untilfurther testing. From the second plate, the fibroblastswere harvested and used for analyses (passage 2). Thepassage 3 fibroblasts were used for RNA harvestingand Northern Blot analysis.

Growth curves. The growth curves wereassessed by plating 1500 passage 2 fibroblasts fromeach cell population in 35-mm culture dishes and bydetermining the total cell number per plate with aCoulter counter on days 7, 10, 13, and 16. The plot-ting of the total cell number against the harvestingdays generated the growth curves. The growth ratewas determined with the slope of a best-fit line thatwas calculated with regression analysis for eachgrowth curve. In addition, the doubling times (DT)were calculated for each cell population. The effect ofb-FGF on the growth rates also was assessed for boththe proximal fibroblasts and the distal fibroblasts at afinal concentration of 100 ng/mL. This concentra-tion was found to have the maximum response test-ed in the previous experiments in our laboratory.2

Assessment of senescence markers in vitro.The fibroblasts were further divided into 3 groups:(1) the fibroblasts that were used for the determina-tion of SA-β–Gal activity (passage 2 fibroblasts), (2)the fibroblasts that were used for the extracellularmatrix protein extraction and the determination offibronectin protein (passage 2 fibroblasts), and (3) thefibroblasts that were used for the extraction of totalRNA and the determination of the fibronectin mes-senger RNA (mRNA) content (passage 3 fibroblasts).

Senescence-associated β-galactosidase activity.The SA-β–Gal staining was performed according to aprotocol that was described previously. Briefly, afterbeing fixed with a 2% formaldehyde/0.2% glutaralde-hyde solution in phosphate-buffered saline solution(PBS), the second passage fibroblast cultures at 30% to50% confluence were incubated with the staining solu-tion (20 mg/mL X-Gal in dimethylformamide; 0.2mol/L citric acid; sodium phosphate buffer, pH = 6.0;100 mmol/L potassium ferrocyanide; 100 mmol/Lpotasium ferricyanide; 5 mol/L sodium chloride; 1mol/L magnesium chloride; and H2O). The plateswere incubated for 12 to 16 hours at 37°C withoutCO2. The positive cells characteristically display a per-inuclear precipitation of blue dye. This allows for a clearidentification with standard light microscopy. The per-centage of positive SA-β–Gal cells is determined after acount of a total of 300 to 400 cells per culture dish.

JOURNAL OF VASCULAR SURGERYVolume 28, Number 6 Mendez et al 1043

exposed to Kodak XAR-5 film. The 28S and the 18Sribosomal RNA subunits that were obtained fromthe ethidium bromide gels were used as controls tonormalize the fibronectin mRNA that was detectedfor each cell population.

Fibronectin probe. The complementary DNAthat was specific for fibronectin was purchased fromthe American Type Cell Culture (Kockville, Md).The complementary DNA was radiolabeled with[32P]-d-CTP (specific activity, 370 MBq/mL) witha commercial oligolabeling kit.

Assessment of senescence markers in vivo. Theimmunohistochemistry analysis of the fibronectindeposition was performed in frozen sections that wereobtained from the skin biopsies procured from theproximal and the ipsilateral distal lower extremities(gaiter area) of the patients with venous reflux. Thespecimens rapidly were frozen and embedded in theoptimal cutting temperature. Thin sections (8 µm)were cut, mounted onto glass slides, and frozen at–70°C until processing. The glass slides then werethawed and washed for 5 minutes with PBS, and thesections were fixed for 10 minutes in 5% formalin, 2mmol/L ethyleneglycol-bis-(β-aminoethylether)-N,N,N’,N’-tetraacetic acid, and 2 mmol/L MgCl2.The slides were incubated at 37°C in the pre-sence of goat serum to reduce nonspecific binding.Subsequently, the slides were incubated with a mouseantihuman cellular fibronectin monoclonal antibody(Sigma) followed by a fluorescein-conjugated rabbitantimouse policlonal antibody (Sigma). A confocal flu-orescent microscope (Leica, Leica Technology, BVRijswijk, The Netherlands) was used to visualize thefibronectin deposition within the dermis of each sec-tion. Laser voltage (990 V) was maintained constantfor each slide, and a 25× objective (0.75 numericalaperture) was used. The images were analyzed withimage analysis software (Image Tools, The Universityof Texas Health Science Center, San Antonio).

Assessment of biopsy site differences infibroblast function. The fibroblasts were isolatedfrom the patients and the volunteers without venousreflux were studied to evaluate whether the site of

the biopsy may influence the fibroblast growth ratesand senescence. The growth rates, the responses tob-FGF, and the SA-β–Gal activity were determinedfor these fibroblasts.

Statistical analyses. A paired Student t test wasused to compare the growth rates, the SA-β–Galactivity, and the density of the fibronectin bands fromimmunoblot analysis. The Mann-Whitney rank sumtest was used to analyze the fibroblasts DT data andthe integrated optical densities that were obtainedfrom the autoradiographs of the Northern blotanalysis. A linear regression model was used to esti-mate the growth rate from each growth curve. Theimage analysis software (Image Tools) was used tocalculate the relative integrated optical densities fromthe scans of immunoblots and Northern blot analy-ses and to calculate the fluorescent intensity that wasobtained from the confocal microscopic images.

RESULTSA total of 23 patients with symptomatic venous

disease were enrolled in this study. At the vascularnoninvasive testing, 18 patients with venous refluxand 5 patients without venous reflux were identified.Of the original 18 patients with reflux, 12 complet-ed all the stages of the study as did 4 of the 5patients without reflux. All of the patients withvenous reflux fell into the symptomatic clinical clas-sification categories 4 and 5, according to theAmerican Venous Forum ad hoc committee consen-sus statement from 1994.10 Only 3 of these patientsfell into category 5. The mean venous filling indexfor the group with venous reflux and the groupwithout venous reflux were 5.45 ± 1.5 mL/s and 1± 0.2 mL/s, respectively. Of the 12 patients withvenous reflux, 4 were men and 8 were women. Themean age of the group was 48.5 years (range, 29 to74 years). No patients had a documented history ofdeep venous thrombosis or major lower extremitytrauma. The associated comorbidity was only hyper-tension in 2 patients.

Fibroblast growth rates in patients withvenous reflux. For the patients with venous reflux,

Table I. Fibroblast growth rates in patients with venous reflux (n = 12)

Growth rates (cells/day) P value

Fibroblast population No b-FGF b-FGF (100 ng/mL)Proximal fibroblasts 29,550 ± 5035 47,030 ± 6133 .012Distal fibroblasts 15,746 ± 4287 41,717 ± 9542 .016P value .016 .53

b-FGF, Basic fibroblast growth factor.

JOURNAL OF VASCULAR SURGERY1044 Mendez et al December 1998

the distal fibroblast growth rates were significantlylower than the proximal fibroblast growth rates in 9of the 12 patients. The mean growth rate for theproximal fibroblasts was 29,550 ± 5035 cells/dayversus 15,746 ± 4287 cells/day for the distal fibro-blasts. A comparison of the proximal fibroblast andthe distal fibroblast growth rates with a 2-tailedpaired t test showed a statistically significant differ-ence (P = .016; Table I). The median DT of theproximal fibroblasts was 43.9 hours versus 66.2hours for the distal fibroblasts (P = .001; Table II).

Senescence-associated β-galactosidase activity.SA-β–Gal activity was detected in 8 of the 12 proxi-mal fibroblast cultures, with a mean of 2.2% ± 0.8%of the cells staining positive. In the distal fibroblastpopulations, SA-β–Gal activity was detected in 12 ofthe 12 patients, with a mean of 8.32% ± 1.9%. Acomparison of the proximal fibroblast versus the dis-tal fibroblast SA-β–Gal activity with a 2-tailed paired

t test showed a statistically significant difference (P =.0084; Table III). No significant correlation betweenthe percentage differences of SA-β–Gal positive cellsand the differences in the growth rates between theproximal fibroblasts and the distal fibroblasts wasfound (data not shown).

Fibronectin protein level and mRNA expres-sion. When comparing the Northern blot analysisbands, the integrated densities were normalized withthe corresponding density of the 18S ribosomal RNAsubunit for each band to account for the differencesin loading. The distal fibroblast fibronectin mRNAwas upregulated in 7 of the 10 patients who were test-ed with a mean normalized integrated density of18.03 ± 4.46 versus 14.3 ± 6.2 of the proximalfibroblasts (P = .013; Fig 1). The distal fibroblastfibronectin protein level was higher in 8 of the 10patients when compared with the correspondingproximal fibroblast cultures with a mean integrated

Fig 1. Fibronectin messenger RNA, as determined with Northern blot analyses, from 80% to90% confluent proximal fibroblast (P) and distal fibroblast (D) cultures of patients with venousreflux but no ulcers. Proximal fibroblasts and distal fibroblasts were cultured at same time withsame culture conditions for each patient before harvest of total RNA. Integrated densities (ID)were calculated from scans of autoradiographs with an image analysis software. Also, ultravio-let light photographs were made of the ethidium bromide gels that showed the 28S and 18Sribosomal RNA subunits. Normalized integrated densities (NID) were obtained with the inte-grated densities of the 18S ribosomal RNA subunit. Fibronectin messenger RNA was upregu-lated in 7 of 10 patients (P = .013).

Fig 2. Fibronectin protein level, as determined by immunoblot technique, from 80% to 90%confluent proximal fibroblast (P) and distal fibroblast (D) cultures of patients with venousreflux but no ulcers. Both cell populations were cultured simultaneously under same cultureconditions for each patient before urea extraction of matrix proteins. An equal amount of totalprotein from proximal and distal fibroblasts was loaded into each lane of sodium dodecyl sul-fate-polyacrylamide gel electrophoresis. Integrated densities (ID) were calculated from scansof autoradiographs with an image analysis software. Fibronectin protein level was upregulatedin 8 of 10 patients (P = .04).

JOURNAL OF VASCULAR SURGERYVolume 28, Number 6 Mendez et al 1045

density of 455.7 ± 80 versus 210 ± 52 (P = .04; Fig2). Six of those 8 patients with increased fibronectinprotein also had increased levels of fibronectin mRNAon Northern blot analysis.

Dermal fibronectin in-vivo (frozen sections).The results of the immunohistochemistry examinationfor fibronectin deposition in the frozen sections areshown in Fig 3. When comparing the proximal dermiswith the distal dermis, a clear increase in the level ofdermal fibronectin deposition was evident in 3 of the6 patients who were tested. This was confirmed by themeasurement of the relative fluorescence intensity withthe image analysis software. When the proximal skinfluorescent intensity was compared with that of thedistal skin with paired t test analysis, no significant dif-ference was noted (P = .87; Table IV).

Effects of b-FGF on growth rates. b-FGF wasmitogenic to both the proximal and the distal fibro-blasts (P < .02). The growth rates for the proximalfibroblasts and the distal fibroblasts in the presence ofb-FGF were not significantly different (47,030 ±6133 cells/day vs 41,717 ± 9542 cells/day; P = .53;Table I). The mean percentage increase in growthrate for the proximal fibroblasts was 35.3% ± 10.6%and for the distal fibroblasts was 45.8% ± 15.2%. This

difference was not statistically significant (P = .34).When the data were analyzed by calculating thefibroblast DT, b-FGF significantly reduced the pop-ulation DT for both fibroblast populations, but thedistal fibroblast DT remained significantly longerthan that of the proximal fibroblasts (Table II).

Fibroblast growth rates in patients withoutvenous reflux. In the subjects without venousreflux, the proximal fibroblast and the distal fibro-blast growth rates, in the presence or absence of b-FGF, were not significantly different (Table V). Thenonstimulated median DTs were 43.9 hours and41.6 hours, respectively (P = .87; Table II). Themean percentage of SA-β–Gal positive cells for theproximal fibroblasts was 1.4% ± 1.05% and for thedistal fibroblasts was 1.7% ± 1.2% (Table III).

Table II. Median doubling times expressed in hours for each fibroblast population

Patients with venous reflux Patients without venous refluxmedian fibroblast DT (hours) median fibroblast DT (hours)

No b-FGF b-FGF (100 ng/mL) P value No b-FGF b-FGF (100 ng/mL) P value

Proximal fibroblasts 43.9 35.7 .001 43.8 39.4 .25Distal fibroblasts 66.2 46.7 .001 41.6 31.6 .25P value .001 .01 .8 .7

Data were compared with nonparametric Mann-Whitney test. On average, the population doubling time of unstimulated distal fibro-blasts exposed to venous hypertension was 2.23-fold longer than proximal fibroblasts.DT, Doubling time; b-FGF, basic fibroblast growth factor.

Table III. Percentage of senescence-associated β-galactosidase activity positive fibroblasts in vitrofrom patients with venous reflux and patients with-out venous reflux

Patients with venous Patients withoutFibroblast population reflux (n = 12) venous reflux (n = 4)

Proximal fibroblasts 2.2 ± 0.8 1.4 ± 1Distal fibroblasts 8.3 ± 1.9 1.7 ± 1.2P value .008 .23

The total number of cells counted by means of light microscopyfrom each cell population was 300 to 400 cells. Subconfluent(40% to 50%) second-passage fibroblasts were used in all patients.

Table IV. Dermal fibronectin protein depositionin biopsies obtained from proximal thigh (proxi-mal) and ipsilateral gaiter area (distal) in 6 patientswith venous reflux

Relative fluorescence intensity

Proximal Distal

72.6 43.1631.76 98.36

127.95 55.91156.24 228.8873.48 119.24

162.7 104.84

Average

104.1 108.3

Fibronectin relative fluorescence intensity was determined fromimmunohistochemical images obtained with a confocal micro-scope with an image analysis software. 25× magnification and alaser voltage of 990 V were kept constant for each slide. Data areexpressed in fluoroscent units and derived from comparisons oftotal fluorescent intensity to a standard curve scanned and usedfor calibration. No significant difference was observed when thedata were analyzed with paired t test (P = .87).

JOURNAL OF VASCULAR SURGERY1046 Mendez et al December 1998

DISCUSSIONFor years, venous stasis ulceration and poor tissue

integrity have been associated with the hydrostatictransmission of deep venous pressure to the fragile,small superficial venous tributaries of the gaiter area.This transmission of pressure occurs in the face ofincompetent valve leaflets whether they are in thesuperficial, deep, or perforating systems.11-13

Numerous pathophysiologic mechanisms havebeen proposed to explain the formation of thevenous ulcers and the nonhealing state that occurs inlimbs that are affected by chronic venous insuffi-ciency. The trapping and activation of white bloodcells,14,15 the effect of the histopathologic findingsof fibrin cuffing,16 and the poor delivery of growthfactors17 have all been explanations that were cham-pioned by various groups.

In our previous studies, the fibroblasts that wereisolated from the venous ulcers displayed reducedgrowth rates and increased senescence characteris-tics.2,3 The hallmark of cellular senescence is an irre-versible arrest of cell proliferation with the mainte-nance of metabolic activity with altered cell func-tions. Cellular senescence depends on cumulative celldivisions and not chronologic or metabolic age,which indicates that proliferation is limited by a“mitotic clock.”19,20 This mitotic clock recently hasbeen characterized to be the length of the chromo-somal telomeres, and the extension of life-span hasbeen achieved through the elongation of the telom-eres after the introduction of telomerase into normalhuman fibroblasts.21 It was not clear in our previousexperience whether the increase of fibroblast senes-cence was related to the intrinsic mechanisms withinthe ulcer environment, such as venous hypertension,or whether the increase was induced by extrinsicmechanisms, such as injury, bacterial colonization, orinfection. In this study, we found that the fibroblaststhat were isolated from intact skin subjected tovenous hypertension have reduced growth rates, havelonger DT, and display more characteristics of cellu-lar senescence than the fibroblasts that were isolatedfrom the skin not exposed to venous hypertension.These data support the concept that (1) the fibrob-last abnormalities that were found in the patientswith venous ulcers preexist in the patients who havevenous hypertension but no ulcers and are the resultof intrinsic mechanisms, and (2) the mechanisms thatare involved act at a local rather than systemic level.

b-FGF was mitogenic for both of the fibroblastpopulations—proximal and distal. This is consistentwith our previous studies.2 Interestingly, b-FGFrestored the distal fibroblast growth rate to that ofthe stimulated proximal fibroblasts. This is in con-trast with our pervious experience in which thegrowth rate of the fibroblasts that were isolated fromthe venous ulcers was restored in most of thepatients to that of the unstimulated normal fibro-blast,2 which implies a higher response rate by thedistal fibroblasts than by the ulcer fibroblasts.

The fibroblasts that migrate from the skin speci-mens to the tissue culture dish by explant techniqueare not senescent by definition because cell prolifer-ation is a requisite. The fibroblast population that issubsequently obtained is heterogeneous on the basisof their in vivo cumulative cell divisions that rangefrom “young” fibroblasts (those that underwentonly a few cell divisions before biopsy) and senescentfibroblasts (those that exhausted their proliferativecapacity). In between, there are fibroblasts at differ-

Fig 3. Representative images of immunohistochemically(IH) localized fibronectin protein deposition within thedermis of patient with venous reflux. Images wereobtained with confocal microscopy at 25× magnification.In image 1 (negative control), mild nonspecific binding ofthe secondary antibody was evident only in epidermis (e).In image 2, fibronectin deposition is localized primarily inthe dermis (d) as expected. Images 3 and 4 show the levelof fibronectin deposition within the dermis of biopsiesobtained from proximal thigh (proximal) and ipsilateralgaiter area (distal) in patient with venous reflux. Anincreased amount of fibronectin strands (▲) can be appre-ciated. With this technique, dermal fibronectin depositionwas found to be increased in the distal dermis of 3 of 6patients tested with venous reflux.

JOURNAL OF VASCULAR SURGERYVolume 28, Number 6 Mendez et al 1047

ent proliferative stages. Senescent or near senescentfibroblasts become resistant to mitogenic stimuli byseveral growth factors.7,22 Then, the b-FGF–mediat-ed increment in the overall growth rate is providedby the proportion of young rapidly proliferatingfibroblasts within a particular cell population. Wecan speculate then that fibroblast populations thatare isolated from intact skin subjected to venoushypertension have a large proportion of young pro-liferating fibroblasts and this is why stimulation withb-FGF restored the growth rate to that of stimulat-ed proximal fibroblasts. We also can speculate thatthe proportion of young fibroblasts within the cellpopulations that are isolated from the patient withvenous hypertension is greater than that of thefibroblasts that are isolated from the venous ulcers,which accounts for the increased responsiveness ofthe former.

The location of the biopsy site could be a poten-tial source of error because, in theory, the distalfibroblasts have undergone a greater number of celldivisions during embryologic development than theproximal fibroblasts. For the subjects with no docu-mented venous reflux, no differences were noted inthe fibroblast growth rates or the senescence bio-markers between the proximal and distal fibroblasts.These findings suggest that the constitutional celldivision difference between the proximal and distalfibroblasts does not seem to contribute significantlywith the changes observed on the distal fibroblastsfrom a patient with venous reflux.

With immunohistochemical analysis, the deposi-tion of fibronectin within the dermis of the skin sub-jected to venous hypertension was found to beincreased in half of the patients studied when com-pared with the normal skin of the same patients.Final dermal fibronectin deposition and stabilizationis a complex process that involves a balance betweenthe fibronectin generated by the fibroblasts, thebreakdown of fibronectin by proteases, and theextravasation of plasma fibronectin. When the fluo-rescent intensities that were measured from eachspecimen were compared, statistical significance was

not reached, which was probably as a result of thesmall number of patients who were tested. Still, thismethod should be considered in the future as apotential biomarker for the in vivo determination ofsenescence characteristics and could be used as aclinical tool to identify the patients with venousreflux who are at risk for the development of venousulcers or to predict the outcome of chronic ulcers.

The mechanisms through which venous hyper-tension, over time, causes dermal fibroblast senes-cence are not known. We can only speculate thatvenous hypertension, through the establishment ofan inflammatory environment, may chronically stim-ulate dermal fibroblast proliferation, which exhauststheir replicative lifespan by shortening the telomerelength. It is also possible that mediators that arereleased locally in this inflammatory environment actas noxious stimuli for susceptible young rapidlyreplicating fibroblasts, which select out for more sta-ble and resistant “old” cells. It is likely that perivas-cular fibrin cuffing, white blood cell trapping andactivation, and growth factor imbalances all playsome role in the selection or accumulation of agedfibroblasts.

The accumulation in vivo of senescent fibroblastsover time is possible because they become resistantto apoptosis on senescence.23 Also, they are unableto proliferate as well as normal fibroblasts,22 are lessmotile,22 and produce a different array of proteins,which includes elevated levels of matrix metallopro-teinases6 and proinflammatory cytokines7 that canaffect tissue integrity and normal healing. This is anovel pathophysiologic concept that may explainsome of the cellular mechanisms of disease involvedin venous ulcer formation and chronic venous insuf-ficiency.

Some potential different therapeutic approachescan be derived from these findings. Bypass ofreplicative senescence has been achieved in the labo-ratory through different mechanisms,21,24 and thistechnology could have therapeutic implications inthe future management of venous ulcers and chron-ic venous insufficiency.

Table V. Fibroblast growth rates in patients without venous reflux (n = 4)

Growth rates (cells/day) P value

Fibroblast population No b-FGF b-FGF (100 ng/mL)Proximal fibroblasts 30,339 ± 9191 52,421 ± 15,746 .27Distal fibroblasts 32,956 ± 10,724 62,964 ± 25,333 .22P value .69 .44

b-FGF, Basic fibroblast growth factor.

JOURNAL OF VASCULAR SURGERY1048 Mendez et al December 1998

CONCLUSIONWe conclude that the fibroblasts that were isolat-

ed from the intact skin subjected to venous hyper-tension exhibited more characteristics of senescencewhen compared with the fibroblasts that were isolat-ed from the unaffected skin of the same patient. Thisimplies that (1) fibroblasts abnormalities exist beforethe development of venous ulcers in patients withvenous reflux and (2) the local accumulation ofsenescent fibroblasts in dermis subjected to venoushypertension or venous ulcers is a result of intrinsicmechanism directly related to venous hypertensionitself and not to extrinsic mechanisms, such as injuryor infection. b-FGF restores fibroblast proliferationin vitro, which suggests a potential therapeutic role.The immunohistochemical determination of dermalfibronectin deposition could be a clinically applica-ble tool to assess senescence characteristics in vivo.

REFERENCES

1. Callum M. Prevalence of chronic leg ulceration and severechronic venous disease in western countries. PhlebologieSuppl 1992;1:6-12.

2. Stanley AC, Park H-Y, Phillips TJ, Russakovsky V, MenzoianJO. Reduced growth of dermal fibroblasts from chronicvenous ulcers can be stimulated with growth factors. J VascSurg 1997;26:994-1001.

3. Mendez MV, Stanley A, Park H-Y, Shon K, Phillips T,Menzoian JO. Fibroblasts cultured from venous ulcers displaycellular characteristics of senescence. J Vasc Surg 1998. In press.

4. Dimri GP, Lee X, Basile G, Acosta M, Scott G, Roskelley C,et al. A biomarker that identifies senescent human cells in cul-ture and in aging skin in vivo. Proc Natl Acad Sci U S A1995;92:9363-7.

5. Murano S, Thweatt R, Shmookler Reis RJ, Jones RA,Moerman EJ, Goldstein S. Diverse gene sequences are over-expressed in Werner syndrome fibroblasts undergoing prema-ture replicative senescence. Mol Cell Biol 1991;11:3905-14.

6. West MD, Pereira-Smith OM, Smith JR. Replicative senes-cence of human skin fibroblasts correlates with loss of regu-lation and overexpression of collagenase activity. Exp Cell Res1989;184:138-47.

7. Campisi J. Replicative senescence: an old lives’ tale? Cell1996;84:497-500.

8. Aggarwal BB, Totpal K, LaPushin R, Chaturvedi MM,Pereira-Smith OM, Smith JR. Diminished responsiveness ofsenescent normal human fibroblasts to TNF-dependent pro-liferation and interleukin production is not due to its effectson the receptors or on the activation of a nuclear factor NF-kappa B. Exp Cell Res 1995;218:381-8.

9. Chandraskhar S, Sorrentino JA, Millis AJ. Fibronectin from

aged fibroblasts is defective in promoting adhesion. J CellPhysiol 1980;103:47-54.

10. Bergan JJ, Eklof B, Kistner RL, Moneta GL, Nicolaides AN.Classification and grading of chronic venous disease in thelower limbs: a consensus statement. In: Gloviczki P, Yao JDT,editors. Handbook of venous disorders: guidelines of theAmerican Venous Forum. 1st ed. New York: Chapman andHall, 1996. p. 652-60.

11. Linton RR. The communicating veins of the lower leg andthe operative technique for their ligation. Ann Surg 1938;107:582-93.

12. Dodd H, Cockett FR. The management of venous ulcers(269-276), the pathology and surgery of the vein of the lowerlimbs, c. 1976. Churchill Livingstone, Edinburgh; 1976. p. 269-76.

13. Linton RR. The post-thrombotic ulceration of the lowerextremity: its etiology and surgical treatment. Ann Surg1953;138:415-30.

14. Thomas RS, Nash, GB, Dormandy JA. White cell accumula-tion in dependent legs of patients with venous hypertension: apossible mechanism for trophic changes in the skin. Br Med J1988;296:1693-5.

15. Pappas PJ, DeFouw DO, Venezio LM, Gorti R, Padberg FTJr, Silva MB Jr, et al. Morphometric assessment of the dermalmicrocirculation in patients with chronic venous insufficien-cy. J Vasc Surg 1997;26:784-95.

16. Burnand KG, Whimster I, Naidoo A, Browse NL. Pericapillaryfibrin in the ulcer-bearing skin of the leg: the cause of lipoder-matosclerosis and venous ulceration. Br Med J 1982;285:1071-2.

17. Falanga V, Eaglstein WH. The trap hypothesis of venousulceration. Lancet 1993;341:1006-8.

18. Hayflick L. The limited in vitro lifetime of human diploid cellstrains. Exp Cell Res 1965;37:614-36.

19. Dell’Orco RT, Mertens JG, Kruse PFJ. Doubling potential,calendar time, and senescence of human diploid cells in cul-ture. Exp Cell Res 1973;77:356-60.

20. Harley CB, Goldstein SJ. Cultured human fibroblasts: distri-bution of cell generation and a critical limit. CellularPhysiology and Biochemistry 1978;97(3 Pt 2 suppl 1):509-16.

21. Bodnar AG, Ouellette M, Frolkis M, Holt SE, Chiu CP, MorinGB, et al. Extension of life span by introduction of telomeraseinto normal human cells. Science 1998;279:349-52.

22. Goldstein S. Replicative senescence: the human fibroblastcomes of age. Science 1990;249:1129-32.

23. Wang E. Senescent human fibroblasts resist programmed celldeath and failure to suppress bcl2 is involved. Cancer Res1995;55:2284-92.

24. Brown JP, Wei W, Sedivy JM. Bypass of senescence after dis-ruption of p21 cip1/waf1 gene in normal diploid humanfibroblasts. Science 1997;277:831-4.

Submitted May 7, 1998; accepted Jul 7, 1998.

Dr Peter J. Pappas (Newark, NJ). Dr Mendez and hiscolleagues at Boston University are to be congratulated ondeveloping an in vitro model to try and determine some of

the pathophysiologic mechanisms associated with chronicvenous insufficiency. In the current study, fibroblasts wereisolated from the gaiter and ipsilateral thigh skin biopsies in

DISCUSSION

JOURNAL OF VASCULAR SURGERYVolume 28, Number 6 Mendez et al 1049

9 patients in class 4, in 3 patients in class 5, and in 4 nor-mal controls. The fibroblast cultures then were analyzedfor growth rate response to basic fibroblast growth factor,fibronectin protein production, and the presence of senes-cence-associated β-galactosidase staining, a marker ofsenescence. The authors noted that the growth rates forunstimulated chronic venous insufficiency fibroblastsshowed decreased growth but were able to proliferate likenormal cells when stimulated with basic fibroblast growthfactor. They also noted no differences in the growth ratesbetween the stimulated and nonstimulated cells that wereobtained from the normal controls. Similarly, they noted ahigher percentage of senescence-associated β-galactosidasecells in the gaiter biopsies of patients with chronic venousinsufficiency as compared with ipsilateral biopsies and biop-sies obtained from the normal controls.

I have several questions for the authors.1. Could you please elaborate for me exactly what the

term senescence means? I know you touched onthis in your presentation. However, to me, senes-cence, or aging, implies an inability to respond tostimuli as a result of the aging process. In your ear-lier publication, you mentioned that the cellsappeared burned out. Could you please clarifyexactly what a senescent cell cannot do functionallythat a young cell can do and what the significance isof fibronectin in determining senescence?

2. The second question regards your study design. Inyour study, you mentioned that 2 of the normal con-trols were patients who had chronic venous insuffi-ciency with no evidence of reflux. Could you elabo-rate on the extent of their venous disease, and doyou think these patients are truly normal controls?

3. Did you compare the growth rates of the distalfibroblasts from patients with chronic venous insuf-ficiency and from normal controls? It appears thatyou only compared the growth rates of the distaland proximal fibroblasts within the same group. Tomy inspection, the growth rates of the patients withchronic venous insufficiency and of the normal con-trols looked different.

4. These data obviously conflict with your earlier pub-lication that showed diminished proliferative growthrates in biopsies obtained from patients with activeulcers. In that earlier publication, cells stimulatedwith growth factors could not even reach prolifera-tive rates of nonstimulated normal controls. In thepresent study, this phenomenon is reversed. Youmentioned in your presentation that this could be aresult of a higher proportion of younger fibroblastsin the cultures that were not senescent. In fact,91.7% of your cells were senescence-associated β-galactosidase negative, and the majority of yourpatients were class 4, as opposed to your previousstudy in which the majority of your patients werepatients with active ulcers.

You also mentioned that the cytochemical environ-ment could affect your results. We at the New Jersey

Medical School would agree with your latter supposition.An electron photomicrograph taken from some of ourpatients shows an intense inflammatory response. You cansee this in a postcapillary venule, with pericapillary cuffing,macrophages that are migrating into lymphatic channels,and the presence of mast cells and macrophages in closeapposition to fibroblasts.

To determine what is going on in the cytochemicalenvironment, we performed immunohistochemistry andlooked for TGF-β–1, and we found that TGF-β-1 waspresent in the epidermis, in the pericapillary cuffing, andin leukocytes. So, we believe that the cytochemical envi-ronment that these cells are exposed to is probably one ofthe more major reasons for some of the events that areoccurring at the cellular level.

In light of these facts, do you think it would be valu-able to stratify your patients according to the clincal, etio-logic, anatomic, and pathologic classification and reana-lyze your data to see if these altered proliferation rates arereally a function of chronic venous insufficiency? Similarly,do you think pretreating your fibroblasts with TGF-β andthen applying your growth factors would be valuable indetermining the mechanisms associated with diminishedgrowth responses?

I greatly enjoyed your study, and I applaud the BostonUniversity group for their interesting and necessary workon fibroblast physiology in patients with venous insuffi-ciency. Thank you very much.

Dr Manuel V. Mendez. Thank you, Dr Pappas, foryour comments and questions.

First, I want to expand on the concept of senescence.Except for a few cells types, there is a limited replicativelife span for all human cells. A senescent stage is reachedafter a certain amount of population doublings. Forfibroblasts, particularly, it is somewhere around 70 to 80cumulative cell divisions.

In the process of reaching senescence in vitro, there isgoing to be a heterogeneous group of fibroblasts. There isgoing to be a proportion of cells that have undergone onlya few divisions—young, rapidly replicating cells that usual-ly respond well to growth factors. There is going to be aproportion of cells that have reached senescence that areunable to divide anymore. At this stage, they cannot repli-cate their DNA despite stimulation by growth factors. Inbetween, there is going to be a group of cells that haveundergone a certain number of population doublings andare near the stage of senescence.

Once the cells get to the senescent stage, as I men-tioned before, they acquire altered cell functions. Therehave been over 200 described altered cell functions acquiredby senescent cells, some of which I alluded in my presenta-tion, like increase in fibronectin production, increase in col-lagenase activity, and decrease in the expression of TIMPs.

Fibronectin is one of the classically described altered cellfunctions that have been demonstrated and used as a bio-marker for the study of senescence in vitro. Interestingly,fibronectin is part of the chemical composition of the fibrincuff that was actually shown in your slide. Also, fibronectin

JOURNAL OF VASCULAR SURGERY1050 Mendez et al December 1998

fragments, when broken down by collagenases or proteases,are chemoattractants to macrophages and actually can acti-vate macrophages. So, you can speculate that an increasedproduction of fibronectin by senescent fibroblasts withinthe skin of patients with venous hypertension may in partexplain why we see leukocyte trapping and maybe activationwithin these tissues.

In terms of choosing our normal controls, we decidedto compare distal fibroblasts versus proximal fibroblastswithin the same patients because we thought that the onlyreal difference between the two cell populations was thatthe distal fibroblast skin had been subjected to venoushypertension. I would not call those patients a “normal”control because, again, these patients do have a chronicvenous insufficiency. However, we thought that thischoice represented the best way to compare these cell pop-ulations. We did not compare fibroblasts between patientsbecause we found that there was a lot of variability in thegrowth rates between patients.

In terms of growth factor effects, we think that infibroblast populations isolated from ulcers, the proportionof younger cells is smaller when compared with thosepatients who do not have an ulcer but do have somevenous hypertension. I think that as time goes by andvenous reflux worsens with the eventual development ofan ulcer, young fibroblasts are pushed to divide andbecome closer to senescence. When we talked about an 8%senescence-associated β-galactosidase positive cell, wewere talking about the population of those fibroblasts thatactually reached senescence in vitro. We were not account-ing for the proportion of fibroblasts that are on their wayto senescence and may not respond well to basic fibroblastgrowth factor. The fact that we are seeing a completeresponse to basic fibroblast growth factor in the venoushypertension patients is probably because that proportionof cells that are closer to senescence is less than the onesobtained from the ulcers.

The question about TGF-β is an interesting one. Weare currently looking at the effects of several proinflam-matory cytokines, like interleukin-1, tumor necrosis fac-tor, and TGF-β, on the function of neonatal foreskinfibroblasts. We have found that the addition of TGF-βactually shows some reduction in the proliferation rate ofthese neonatal fibroblasts. If you review the literature,TGF-β–1 can have either stimulatory effects or inhibitoryeffects on the proliferative capacity, depending on the con-centration used for tissue culture. I do not know whetherthe addition of TGF-β to our cells would modify theresponse to basic fibroblast growth factor.

In terms of stratifying the data, unfortunately, we didnot have a large enough group of patients to stratify andanalyze the data on the basis of the classification used inthe clinical, etiologic, anatomic, pathologic system.

Dr Robert E. Madden (Valhalla, NY). I was just won-dering if you have looked at or you plan to do any look-ing for telomerase activity in your tissue cultures?

Dr Mendez. Recently, interesting experiments haveshown that the length of the telomeric DNA acts as the

“mitotic clock” that limits replicative capacity. Every timea cell divides, the telomere length shortens. And it gets toa point that beyond a certain length the cells would notundergo replication again. We have thought about mea-suring telomere length of proximal and distal fibroblasts.Telomerase has been successfully transferred in normalcells to prolong or lengthen the telomere to overcomesenescence.

Frank T. Padberg, Jr (Boston, Mass). I have a plearegarding terminology. You used the terms chronic venousinsufficiency, venous hypertension, and reflux inter-changeably, but you have measured reflux. You imply thatreflux leads to venous hypertension; however, other mea-surements (residual volume function, or RUF) availablethrough your testing apparatus, air plethysmography, cor-relate more closely with venous hypertension as deter-mined by venous pressure studies. Could you comment onthat, please?

Dr Mendez. We use air plethysmography as our prima-ry inclusion criteria for the patients. We chose a venousfilling index of greater than 2 mL/s to define venoushypertension. We also used duplex scan examination tolook at the deep, superficial, and perforated system to cor-roborate or to confirm the presence of venous reflux.

Dr Anton N. Sidawy (Boston, Mass). When you useda basic fibroblast growth factor, did you use it on thesenescent cells and the nonsenescent cells? And if you did,was there a difference in the nonsenescent cells before andafter basic fibroblast growth factor treatment?

Dr Mendez. We did use basic fibroblast growth factoron all fibroblast populations. Again, the fibroblast popula-tion in vitro is a heterogenous group of fibroblasts. Youare going to find senescent and nonsenescent cells. Welooked at the growth rates without basic fibroblast growthfactor, which is when we saw that there was a significantdifference between the proximal and distal fibroblasts(30,000 cells/day vs 15,000 cells/day). When we addedbasic fibroblast growth factor, we first observed a signifi-cant increase in the growth rates of both proximal and dis-tal fibroblasts.

It appeared that the response of the distal fibroblastwas higher than the response of the proximal fibroblast.But on statistical analysis, there was no significant differ-ence. We found that adding basic fibroblast growth factorbrought up the growth rate of the distal fibroblasts to thatof stimulated proximal fibroblasts. The increment ingrowth rate is a result of the response to basic fibroblastgrowth factor by the proportion of young fibroblasts with-in these cell populations.

Dr Sidawy. Did you do the significance on the effect ofno basic fibroblast growth factor and the basic fibroblastgrowth factor on the proximal cells?

Dr Mendez. Yes. When we looked at these 2 cells andcompared, there was a significant response to basic fibro-blast growth factor. And again, it was significant in the dis-tal fibroblasts. When we looked at the difference betweenthe 2 of them with basic fibroblast growth factor, therewas no significant difference.