Preferential vascular-based transfer from vagina to the corpus but not to the tubal part of the uterus in postmenopausal women

Niels Einer-Jensen1,3, Ettore Cicinelli2, Pietro Galantino2, Vincenzo Pinto2 and Bruno Barba2

1 Physiology and Pharmacology, University of Southern Denmark, Winsloewparken 21, DK-5000 Odense M, Denmark and 2 Department of Obstetric and Gynaecology, University of Bari, Policlinico, Piazza Giulio Cesare, 70124 Bari, Italy


    Abstract
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
BACKGROUND: Vaginal administration of progesterone during infertility treatment has therapeutic advantages over oral administration. However, the reasons for this are poorly defined. To demonstrate a preferential vagina-to-uterus distribution of substances, we investigated cold distribution from vagina to the uterus and rectum. METHOD: In 10 postmenopausal women, thermoprobes were inserted into the uterine cavity and in the rectum at <9 cm or at >9 cm from the anus; temperatures were subsequently measured during 10 min flushing of vagina with cold saline. RESULTS: After 10 min, temperature decreased as follows: uterus, tubal angle: –0.22 ± 0.07°C, 10 (mean ± SEM, n); uterus, middle cavity: –1.26 ± 0.34°C, 9; rectum, <9 cm insertion: –3.69 ± 0.68°C, 3; rectum, >9 cm insertion: –0.51 ± 0.19°C, 6. CONCLUSIONS: Despite obviously different distances to the vagina of the uterine and the low rectal probes (<9 cm) the temperature decrease occurred at the same time. Cold transfer from vagina to the uterus and rectum is probably not the result of simple diffusion but of a vascular counter-current transfer. Differential cooling of corpus and tubal angles suggests a different arterial supply; while uterine corpus is supplied from the uterine artery, the tubal angles seem to be mainly supplied from the ovarian artery via the tubal arcade.

Key words: postmenopausal women/tubal vascularization/uterine artery/uterine vascularization/vagina-to-uterus


    Introduction
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
Vaginal administration of progesterone in infertility treatment has led to therapeutic advantages over the use of oral synthetic progestins. Unexpectedly, a wealth of evidence soon accumulated indicating that the effects on the uterus of the vaginal progesterone exceed the expectations reasonably drawn from the relatively low blood concentrations achieved (De Ziegler, 1995Go; Fanchin et al., 1997Go). The discrepancy between the uterine effects that mimic menstrual cycle findings and blood concentrations in the luteal phase defect range has led to an hypothesis suggesting preferential delivery of progesterone to the uterus, or `first uterine pass effect' (De Ziegler, 1995Go). Puzzled by these findings, several authors have attempted to confirm the existence of direct vagina-to-uterus transport by documenting that vaginal administration results in high concentrations of progesterone in uterine tissue despite low concentrations in peripheral blood (Miles et al., 1994Go; Cicinelli et al., 2000Go).

Contrasting with the mounting evidence that now supports the existence of a first uterine pass effect, the mechanism by which this functional `portal' system linking the vagina-to-uterus is mediated remains a matter of debate. In previous studies, we showed that progesterone administered vaginally resulted in concentrations in the uterine artery that largely exceeded that of peripheral arteries in pigs (Einer-Jensen et al., 1993Go) and humans (Cicinelli et al., 1998Go). This finding could not be explained as progesterone being absorbed from the vaginal mucosa and distributed via the peripheral circulation. Hence, this unexpected observation was the first clue that led to the proposal of a counter-current mode of exchange with an upward vagina-to-uterus transport of progesterone absorbed in vaginal and lymphatic vessels and diffusing to nearby arteries (Cicinelli and De Ziegler, 1999Go). Counter-current vein-to-artery exchanges may take place each time a difference in concentration exists between fluids circulating in opposite directions in nearby vessels and sharing a large exchange surface. In the kidney, counter-current exchanges occurring between the descending and ascending arms of the Henle's loop affect chemical substances (electrolytes). In some areas of the body, counter-current exchange results in the transfer of heat. For example, transfer takes place between the testicular veins and artery and in the sinus cavernosus. The venous effluent from the nasal area, cooled by outside air, serves to lower the temperature of blood ascending in the carotid artery, thus keeping the brain cool (Baker, 1979Go; Glad Sorensen et al., 1991Go).

As the nature of counter-current exchange of substances is similar to the transfer of heat, we decided to study further the local transport linking the vagina to the uterus with an experimental in-vivo model that looks at heat exchange between the vagina and the uterus. Isotonic saline at room temperature was instilled into the vagina of volunteers while the temperature was recorded in both nearby (low rectum) and distant areas (middle uterus, tubal corner of uterus, and high rectum).


    Materials and methods
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
The investigation was performed at the Menopausal Clinic, University Hospital in Bari, Italy. The local ethical committee approved the study protocol. Ten healthy menopausal women who had never received hormone replacement therapy participated after giving informed consent; their age ranged from 53 to 64 years (56.80 ± 3.97, mean ± SD), the years of menopause from 3 to 11 years (6.10 ± 2.88) and value of body mass index (weight/height2) (kg/m2) between 21 and 25 (23.65 ± 1.28). All were parous women who needed endometrial biopsy because of hysteroscopic diagnosis of low-risk endometrial hyperplasia. None reported symptoms related to or presented signs of vaginal dystrophy at clinical examination. Before performing endometrial biopsy, a thermoprobe was inserted into both the uterine cavity and rectum. The women were conscious during the investigation; no sedative or painkillers were administered. The trial followed safe gynaecological procedures; the temperature probes were autoclaved before use. Careful disinfection of external and internal genitalia preceded the insertion of probes.

Thermoprobes based on Cu/CuNi elements were used. A stainless steel uterine probe (ELLAB, Roedovre, Denmark (www.ELLAB.com), length 25 cm, and outer diameter (OD) 2.5 mm, blunt tip) was developed for the trial. The probe resembled a traditional uterine sound carrying a centimetre scale in order to give hysterometric evaluation; the probe was slightly curved 3–5 cm from the tip. Two thermoelements were built in: in the tip and 3 cm from the tip respectively. The probe was inserted inside the uterine cavity until the tip was in contact with one of the uterine cornua so that consequently the lower thermoelement was in the middle of the cavity (Figure 1Go). Correct placement of the uterine probe was assessed by ultrasound and care was taken to maintain the same position of the probe throughout all the experiment. A 5 mm diameter thermoprobe (ELLAB, MRC55044A) was also inserted into the rectum as a reference. The depth of insertion varied from 5 to 15 cm. Anatomically, the length of the posterior vaginal wall is estimated to vary between 8 and 10 cm (Williams et al., 1989Go). The results of the rectal measurements were therefore allocated into two groups: <9 cm (low rectum) insertion and >9 cm (high rectum) insertion of the probe. In the first group, it was possible to feel the tip of the thermoprobe through the posterior vaginal wall during the gynaecological examination. In the other group, the probe was placed deeper so that at vaginal examination the tip of the probe was higher than the posterior vaginal fornix. In one case, the probe was moved from position <9 cm to >9 cm during the cooling period to obtain comparable values from the same volunteer. The position of the different measurement sites is shown in Figure 1Go.



View larger version (48K):
[in this window]
[in a new window]
 
Figure 1. The pelvic organs in a woman. The arrows indicate the position of temperature measurement sites in the uterine cavity (tubal corner and corpus of uterus) and rectum (probe inserted <9 cm and >9 cm).

 
One minute after insertion of the probes, the vagina was flushed for 10 min with sterile saline (1000–1500 ml per 10 min, 22°C). The probes were removed after stopping the flushing.

All temperature tracings were printed and the changes evaluated visually. The temperature readings at baseline, 2 and 10 min after starting of the vaginal flushing were measured manually on each graph. Two minute readings were selected to catch `rapid' changes and 10 min chosen as the `steady state' value. The change between baseline, 2 and 10 min temperatures was calculated. Statistical evaluation was performed by paired and unpaired Student's t-tests as appropriate; P < 0.05 was considered significant.


    Results
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
Intrauterine recordings were obtained in 10 patients, and rectal recordings in nine patients (five cases >9 cm, four cases <9 cm of insertion). Three women experienced slight and brief pain during passage of the cervical canal; the rest of the experiment was painless for all women. At hysterometry, the length of the uterine cavity varied between 7.5 and 8.5 cm.

Changes in the uterine and rectal temperatures are illustrated in Figure 2Go. The decrease in organ temperature did not follow parallel patterns. After 10 min flushing the decrease in organ temperatures was as follows: uterus, tubal angle: –0.22 ± 0.07°C, 10 (mean ± SEM, n); uterus, middle cavity: –1.26 ± 0.34°C, 9 (in one patient a huge decline in temperature occurred and she was considered an outlier and removed from calculation; see Figure 3Go); rectum, <9 cm insertion: –3.69 ± 0.68°C, 4; rectum, >9 cm insertion: –0.51 ± 0.19°C, 5. The temperature decrease in the uterine body and in the lower part of the rectum was fast and pronounced; the two other measurement sites showed a delayed and much less pronounced cooling. The small changes in temperature observed in the high rectum (>9 cm insertion) and in the tubal angle of the uterine cavity occurred gradually during the 10 min flushing period. In contrast to this, the larger decreases (>1.0°C) in temperatures were rapid (<2 min) in the middle of the uterine cavity and in the low rectum (Figure 2Go). The middle of the uterine cavity was cooled significantly more and faster than tubal angles (P = 0.015 at 2 min and 0.004 at 10 min, paired t-test) (Figures 2 and 3GoGo). Similarly, cooling was faster and more pronounced in the rectum when measured <9 cm from the anus compared with the average decrease in temperature measured >9 cm from the anus (P = 0.187 at 2 min and 0.0005 at 10 min, non-paired t-test). The temperature decrease in the rectum was not linearly correlated with the distance from the vagina; moving the probe just 2 cm induced a sudden temperature change (Figure 4Go).



View larger version (13K):
[in this window]
[in a new window]
 
Figure 2. Temperature changes in uterus (corpus and tubal part) and rectum (<9 cm and >9 cm insertion) 2 and 10 min after start of vaginal flushing with saline. Values are mean – SEM.

 


View larger version (19K):
[in this window]
[in a new window]
 
Figure 3. Comparison between uterine temperatures before and after 10 min vaginal flushing with saline. The individual values are shown.

 


View larger version (31K):
[in this window]
[in a new window]
 
Figure 4. Temperature changes measured in one patient before, during, and after vaginal flushing with saline. The temperature in the corpus of uterus decreased 1°C during the flushing, while the decrease in the tubal part was small. The rectal temperature (probe inserted 8 cm) decreased also after a latency period. When the rectal probe was moved 2 cm deeper, a pronounced increase in temperature was seen. Full recovery took more than 10 min.

 
The cooling was partly reversed when collection of data was continued for some minutes after stopping the flushing.


    Discussion
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
Many traditional concepts on basic regulatory mechanisms involved in reproductive physiology have been revised recently. Nowadays, experimental and clinical evidences speak for the existence in the female pelvis and for their relevant functional role of local regulatory mechanisms based on preferential transfer of substances and hormones; the transfer may have functional implications (Einer-Jensen and Hunter, 2000Go). The existence of a local functional interplay between ovary, oviduct and uterus in women has been supported by bilateral ultrasonographic Doppler flow studies of the uterine arteries and of arterial anastomoses between the uterine and ovarian arteries (junctional vessels) in the cornual region of both sides of the uterus during the menstrual phase of regular-cycling women. These studies demonstrated significant lower resistance indices of the uterine artery and of the junctional vessels ipsilateral to the side of the dominant ovarian structure as compared with the corresponding arteries contralaterally (Santolaya-Forgas, 1992Go; Kunz et al., 1998Go). Accordingly, vascular perfusion of the fundal myometrium was found to be significantly increased ipsilaterally to the dominant follicle during the late follicular phase of the cycle (Kunz et al., 1998Go). Therefore, ovarian control over uterine function is thought today to be not only exerted via the systemic circulation but also directly, most probably utilizing the utero-ovarian counter-current system (Einer-Jensen, 1988Go; Krzymowski, 1992Go; Kunz et al., 1998Go).

A preferential transfer from vagina to the uterus has also been suggested. Vaginal application of progesterone delivered relatively more hormone to the uterus than i.m. administration (Miles et al., 1994Go; Cicinelli et al., 2000Go). However, the physiological mechanisms involved in this phenomenon phrased `uterine first pass effect' are still not fully understood. Among the different mechanisms advocated, a counter-current exchange between adjacent veins and arteries may play a relevant role (Cicinelli and De Ziegler, 1999Go). In the present study, we found that both uterine and rectal temperatures decreased when the vagina was flushed for 10 min with saline at room temperature. However, we did not observe a simple correlation between the amplitude of temperature changes and the distance of temperature recordings from the vagina. In fact, despite obvious different distances between vagina and the uterine and low rectal probes (group <9 cm), the temperature decrease occurred at the same time. Conversely, at roughly equal distance from the cooled vagina, temperature changes were markedly different in the uterus and high rectum (group >9 cm). Moreover, the temperature in the middle of the uterine cavity (4–6 cm from the vagina) dropped fast and significantly, while the temperature in the tubal part, only a further 2.5–3 cm upper into the uterine cavity, decreased slowly and marginally. These observations indicate that cooling is not the result of simple diffusion but probably of preferential distribution based on a counter-current transfer. Anatomically, the vaginal artery, often found as double or triple, originates from the internal iliac artery, runs along the ipsilateral side of the vagina and communicates through anastomoses with the contralateral vessel(s). In addition, it sends branches to urethra and the fundal area of the bladder, as well as branches to the rectum along the entire posterior vaginal wall (Williams et al., 1989Go). Moreover, the uterine artery forms an arcade with the ovarian artery along the Fallopian tube; this arterial arcade supplies the Fallopian tube and in animals the tubal part of the uterine horn (Schramm et al., 1986Go). It is uncertain where the functional border between the blood supply from the two vessels is located in humans. Most of the vaginal venous outflow reaches the hypogastric veins. Part of the outflow reaches other larger veins via communicants with the uterine, the vesical and the haemorrhoideal venous plexi; the veins also exchange branches with the rectal veins along the entire posterior vaginal wall. The veins draining the vagina flow also communicate with the contralateral veins. Therefore, it seems evident that all routes for the vaginal blood outlet are in close contact with arteries supplying the uterus, the last part of the rectum and, probably, the urethra and part of the bladder (Williams et al., 1989Go). The vaginal wall, and therefore the venous blood, is cooled by the saline. Due to a close contact between the veins and arteries, the colder vein blood will cool the arterial blood. The non-vaginal tissue supplied with the cooler arterial blood will therefore decrease its temperature. The mechanism is similar to the heat transfer taking place between the pampiniform venous plexus and the testicular artery in the boar (Waites and Moule, 1961Go). Temperature gradients between Graafian follicles and ovarian stroma also indicate heat transfer in the ovarian adnexa and between the intra-ovarian blood vessels (Hunter et al., 2000Go).

Major and rapid changes in temperature were found in the corpus of the uterus. We postulate that such changes happen because the corpus is supplied with cooled arterial blood from the uterine artery. The rapid onset indicates that vaginal flushing does not create a gradual change in temperature due to transmission through cervical tissues. The rapid cooling detected by the lower intrauterine thermoprobe makes the possibility of diffusion of saline through the cervical canal less likely. Additionally, if a plastic tube containing water at room temperature was inserted into the vagina, a similar cooling problem was found (N.Einer-Jensen and E.Cicinelli, personal communication).

The small or absent cooling in the Fallopian angle of the uterine cavity indicates that in the cases investigated the arterial blood does not originate from the uterine artery, but is supplied by the ovarian artery through the utero-ovarian arcade. Whether vascular supply of tubes and cornual part of the uterus depends on the uterine or ovarian artery is still debated. Arteriographic investigations performed in women several decades ago suggested that the uterine artery supplies the medial two-thirds of the tube while the ovarian artery supplies only the external third (Borell and Fernstrom, 1953Go). However, the authors reported a wide intersubject variability and technical difficulties in performing the procedure, so that ~50% of cases enrolled were excluded from the evaluation (Borell and Fernstrom, 1953Go). These conclusions disagree with more recent investigations performed in animals. The direction of blood flow in the utero-ovarian anastomosis was studied by direct visual observation in anaesthetised guinea pigs; blood flowed towards the uterus in all observations on all guinea pigs studied (Hossain and O'Shea, 1983Go). Accordingly, in rabbits perfused with methylmethacrylates it has been demonstrated that ovarian artery provides the main blood supply of the ovary, the tube and the tubal part of the uterine horn (Riedel and Cordts-Kleinwort, 1988Go).

Although our findings should be confirmed in premenopausal women, it is conceivable that the tubal part of the uterus may be part of the local transfer system of steroids and peptides in the ovarian adnexa involving the ovarian and tubal blood and lymph vessels (Schramm et al., 1986Go; Einer-Jensen, 1988Go). This opens the possibility that a developing embryo may communicate with the corpus luteum by local hormonal means even after its arrival in the uterus (Hunter et al., 2000Go).

The efficacy of 133xenon transfer from vagina to uterus in rats was influenced by the oestrous cycle (Zhao and Einer-Jensen, 1998Go). Similarly, the stage in the oestrous cycle influenced the transfer of steroid (J.Skipor and N.Einer-Jensen, personal communication) and peptide hormones between blood vessels in the head of pig and sheep (Grzegorzewski et al., 1997Go). The present investigation involved menopausal women (for 3 to 11 years) who had never received hormone replacement therapy and who were not complaining of vaginal dystrophy, in order to obtain a group of patients as homogeneous as possible concerning endocrine milieu and vaginal mucosa trophicity. Further studies are needed to evaluate whether the ovulatory cycle or hormonal treatment influences transfer from the vagina.

A correlation was found between the rectal anatomy and the degree of cooling. The last 9 cm of the rectum cooled very rapidly. We postulated that this is due to the close anatomical connection between vagina and the last part of the rectum, and/or heat transfer between vaginal venous outlet and the rectal arteries. The method could not distinguish between the two possibilities. When the rectal probe was moved >9 cm inside the anus, the temperature decrease was only 1.30 ± 0.51°C compared with 4.77 ± 0.84°C when the distance was <9 cm; the variation happened step-wise as shown in Figure 4Go.

In conclusion, our results indicate that a preferential vascular-based distribution from vagina to the uterus occurs supporting the so-called hypothesis of the first uterine pass effect after vaginal administration (Miles et al., 1994Go; Fanchin et al., 1997Go; Cicinelli et al., 2000Go). Further, although a confirmation in premenopausal women is needed, our data suggest that functionally the arterial blood supply to the tubal part of the uterus originates from the ovarian artery, and not from the uterine artery. The concept that the tubal angle of the uterine cavity receives hormonal stimuli directly from the corpus luteum may open new perspectives in assisted reproduction.


    Acknowledgements
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
We gratefully acknowledge the financial support from our universities, NOVO-Nordisk, Italy, and Wyeth Lederle, Italy.


    Notes
 
3 To whom correspondence should be addressed. E-mail: n.einer-jensen{at}imbmed.sdu.dk Back


    References
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
Baker, M.A. (1979) A brain-cooling system in mammals. Sci. Am., 240, 130–139.[ISI][Medline]

Borell, U. and Fernström, I. (1953) The adnexal branches of the uterine artery: an arteriographic study in human subjects. Acta Radiol., 40, 561–582.[ISI][Medline]

Cicinelli, E. and De Ziegler, D. (1999) Transvaginal progesterone: evidence for a new functional `portal system' flowing from the vagina to the uterus. Hum. Reprod. Update, 5, 365–372.[Abstract/Free Full Text]

Cicinelli, E., Cignarelli, M., Sabatelli, S. et al. (1998) Plasma concentrations of progesterone are higher in the uterine artery than in the radial artery after vaginal administration of micronized progesterone in an oil-based solution to postmenopausal women. Fertil. Steril., 69, 471–473.[ISI][Medline]

Cicinelli, E., De Ziegler, D., Bulletti, C. et al. (2000) Direct transport of progesterone from vagina to uterus. Obstet. Gynecol., 95, 403–406.[Abstract/Free Full Text]

De Ziegler, D. (1995) Hormonal control of endometrial receptivity. Hum. Reprod., 10, 4–7.[ISI][Medline]

Einer-Jensen, N. (1988) Counter-current transfer in the ovarian pedicle and its physiological implications. Oxf. Rev. Reprod. Biol., 10, 348–381.[ISI][Medline]

Einer-Jensen, N. and Hunter, R.H. (2000) Physiological and pharmacological aspects of local transfer of substances in the ovarian adnexa in women. Hum. Reprod. Update, 6, 132–138.[Abstract/Free Full Text]

Einer-Jensen, N., Kotwica, J., Krzymowski, T. et al. (1993) Rapid absorption and local redistribution after vaginal application of progesterone in gilts. Acta Vet. Scand., 34, 1–7.[ISI][Medline]

Fanchin, R., De Ziegler, D., Bergeron, C. et al. (1997) Transvaginal administration of progesterone. Obstet. Gynecol., 90, 396–401.[Abstract/Free Full Text]

Glad Sorensen, H., Lambrechtsen, J. and Einer-Jensen, N. (1991) Efficiency of the counter current transfer of heat and 133xenon between the pampiniform plexus and testicular artery of the bull under in vitro conditions. Int. J. Androl., 14, 232–240.[ISI][Medline]

Grzegorzewski, W.J., Skipor, J., Wasowska, B. and Krzymowski, T. (1997) Counter current transfer of 125I-LHRH in the perihypophyseal cavernous sinus-carotid Rete vascular complex, demonstrated on isolated pig heads perfused with autologous blood. Domest. Anim. Endocrinol., 14, 149–160.[ISI][Medline]

Hossain, M.I. and O'Shea, J.D. (1983) The vascular anatomy of the ovary and the relative contribution of the ovarian and uterine arteries to the blood supply of the ovary in the guinea pig. J. Anat., 137, 457–466.[ISI][Medline]

Hunter, R.H.F., Bögh, I.B., Einer-Jensen, N. et al. (2000) Intra-ovarian temperature gradients in mammals: pre-ovulatory Graafian follicles are cooler than neighbouring stroma. Hum. Reprod., 15, 273–283.[Abstract/Free Full Text]

Krzymowski, T. (1992) New pathways in animal reproductive physiology frontiers and perspectives. J. Physiol. Pharmacol., 43, 5–19.[Medline]

Kunz, G., Herbertz, M., Noe, M. and Leyendecker, G. (1998) Sonographic evidence for the involvement of the utero-ovarian counter-current system in the ovarian control of directed uterine sperm transport. Hum. Reprod. Update, 4, 667–672.[Abstract/Free Full Text]

Miles, R.A., Paulson, R.J., Lobo, R.A. et al. (1994) Pharmacokinetics and endometrial tissue levels of progesterone after administration by intramuscular and vaginal routes: a comparative study. Fertil. Steril., 62, 485–490.[ISI][Medline]

Riedel, H.H. and Cordts-Kleinwort, E.G. (1988) Vascular imaging in the area of the uterine horn and adnexa in the rabbit. Zentralbl. Gynakol., 110, 232–240.[ISI][Medline]

Santolaya-Forgas, J. (1992) Physiology of the menstrual cycle by ultrasonography. J. Ultrasound Med., 11, 139–142.[Abstract]

Schramm, W., Einer-Jensen, N., Schramm, G. and McCracken, J.A. (1986) Local exchange of oxytocin from the ovarian vein to the ovarian arteries in the sheep. Biol. Reprod., 34, 671–680.[Abstract]

Waites, G.M.H. and Moule, G.R., (1961) Relation of vascular heat exchange to temperature regulation in the testis of the ram. J. Reprod. Fertil., 2, 213–224.[ISI]

Williams, P.L., Warwick, R., Dyson, H. and Bannister, L.H. (editors) (1989) Gray's Anatomy. Churchill Livingston, Edinburgh, pp. 1847–1880.

Zhao, J. and Einer-Jensen, N. (1998) Cycle dependent variations in transfer of 133xenon from vagina to uterus in rats. Maturitas, 28, 267–270.[ISI][Medline]

Submitted on November 1, 2000; accepted on March 20, 2001.