Low urine osmolarity as a determinant of cisplatin-induced nephrotoxicity

Authors


Abstract

Cisplatin is widely used in the treatment of human tumors, but it is a nephrotoxic drug. Early pragmatic clinical trials have shown that cisplatin-induced renal toxicity is greatly reduced through the use of high hydration, a large NaCl supply and mannitol infusion, but the precise mechanisms of these nephroprotective measures are not fully understood. We show here an increase in the cisplatin uptake and cytotoxicity on 56/10 A1 human glomerular and HK-2 human tubular cells when the drug incubation was performed in a hypotonic phosphate-buffered saline solution or in human urine (“drag in” transport hypothesis). When 4 mg/kg cisplatin was intraperitoneally injected in rats in 20 ml of a hypotonic 4 g/l NaCl solution, the platinum accumulation increased in both the cortex and papilla but not in the subcutaneously grafted colon tumors when compared to rats injected with cisplatin in normal or hyperosmotic solutions (9 and 14 g/l NaCl, respectively). The urea and creatinine blood levels were significantly increased, and more apoptotic cells were detected by the caspase-3 cleavage and TUNEL assays in the tubular cells of rats treated with cisplatin in a hypotonic solution compared to animals that received normal or hypertonic solutions. Osmolarity was sometimes low in urine from patients receiving an intravenous hydration for a cisplatin treatment or from healthy volunteers who were given an oral hydration with a 50 g/l glucose solution. Our results show that low urine osmolarity could be a major determinant in the increase of cisplatin-induced nephrotoxicity and justify the widely used concurrent infusion of osmotically active substances during intravenous hydration. © 2004 Wiley-Liss, Inc.

Cisplatin is a heavy metal compound with wide antineoplastic activity on a variety of solid malignancies. It cannot be replaced by the less toxic carboplatin for the treatment of many tumors, including germ cell tumors.1, 2 Cisplatin is mainly excreted by the kidneys. Acute and cumulative renal toxicity are the major limitations for its prolonged use.3, 4 Little is known about the definite mechanisms of cisplatin-induced nephrotoxicity. Individual factors, such as platinum plasma levels, turned out to be ineffective or routinely lacking in predicting and preventing accidental episodes of acute renal failure.5 Large prehydration and concomitant osmotic diuresis with mannitol have been demonstrated to prevent most of the nephrotoxicity in dog and man6 and have permitted the wide clinical use of cisplatin.7, 8, 9 Nephroprotection must be maximal when cisplatin is used in a curative intent against germ cell tumors since most of these young men are long-term survivors. Although acute renal failure is now a rare event, chronic alterations of kidney function are frequent. Reduction in the glomerular filtration rate occurs in 20–30% of patients despite intensive prophylactic hydration and forced diuresis.10, 11 Such changes in renal function are essentially irreversible.12 Moreover, 25–39% of patients developed diastolic arterial hypertension after chemotherapy, which was partly related to the renal toxicity.13, 14 A better knowledge of factors that govern cisplatin nephrotoxicity remains an actual problem and could result in an improvement of nephroprotective measures.

Low osmolarity of the culture medium markedly increases the cisplatin accumulation and cytotoxicity in vitro on a variety of cancer cell lines.15, 16, 17 This enhanced penetration was explained by a ‘solvent drag effect’ mechanism of cisplatin passively entering the cells after the movement of water molecules through the plasma membrane. Casual observation leads us to investigate the role of low osmolarity on cisplatin nephrotoxicity. Intraperitoneal injection of 4 mg/kg cisplatin led to the death of rats in a few days when the drug was diluted in 20 ml of a 4.5 g/l NaCl solution (154 mOsm/l), whereas animals that received the same dose in 20 ml of a 9 g/l NaCl solution (308 mOsm/l) survived. Death was explained by acute renal failure. We hypothesized here that a reduced osmolarity in the renal glomerules and tubules could enhance cisplatin accumulation and thus its nephrotoxicity. We investigated the effect of osmolarity on cisplatin accumulation and toxicity (i)in vitro on normal human tubular and glomerular cells and (ii)in vivo on rat kidneys and subcutaneously grafted colon tumors.

MATERIAL AND METHODS

Cells

HK-2 (human kidney-2) is an immortalized proximal tubular cell line from a normal human kidney cortex that was obtained from the American Type Culture Collection (Manassas, VA). HK-2 cells have a brush-border and express alkaline phosphatase, cytokeratin, fibronectin, β1 integrin, glucose transporter, C3a, AQP1 and MDR.18 HK-2 cells were grown in Keratinocyte-Serum Free Medium with 5 ng/ml human recombinant EGF and 0.05 mg/ml bovine pituitary extract (GIBCO, Grand Island, NY) and exposed to 5% CO2 in air atmosphere. 56/10 A1 is a T-SV-40 immortalized human renal epithelial cell line of glomerular origin that retains most of the phenotypic features (expression of cytokeratin, AQP1, CALLA, PHM, t-PA, uPA and PAI.1) of parental glomerular visceral epithelial cells.19 The culture medium was a mixture (50/50, v/v) of Dulbecco's and Ham's F10 medium supplemented with 10% fetal bovine serum.

Cellular accumulation of platinum in vitro

For the platinum accumulation assay, renal cells were seeded for 72 hr in 6-well plates until confluent, then incubated for 3 hr with 20 μg/ml cisplatin diluted in a phosphate-buffered saline solution (PBS, pH 7.4) or urine from a healthy volunteer. The PBS osmolarity was adjusted to 150 mOsm/l (hypotonic), 300 mOsm/l (isotonic) or 600 mOsm/l (hypertonic). High urine osmolarity resulted from a 12 hr drinking abstinence, whereas low osmolarity followed absorption of 2 liters of water in 3 hr. During this restriction-dilution sequence, urine osmolarity varied from 861 to 133 mOsm/l. At the end of drug incubation with PBS or urine, cells were washed twice with isotonic PBS, then detached with a trypsin/EDTA solution and centrifuged. Cell pellets were lyophilized. After dilution in distilled water, platinum was assayed by Atomic Absorption Spectrometry (AAS) in a Zeeman graphite furnace atomic absorption spectrometer (Spectr AA-220 Z Varian, Les Ulis, France) at 266 nm.

In vitro cytotoxicity of cisplatin on renal cells

56/10 A1 cells and HK-2 cells were detached from culture flasks by trypsin/EDTA and seeded (20 × 104 cells/well in 200 μl of culture medium) in 96-well tissue culture plates. After 72 hr of culture, confluent cells were incubated for 3 hr with cisplatin diluted in 200 μl of PBS with various osmolarities. Cells were rinsed twice with isotonic PBS and then recultivated for 72 hr in drug-free culture medium. Cell survival was measured by a Crystal Violet colorimetric assay. Surviving adherent cells were fixed for 5 min with pure ethanol and stained for 10 min with 1% Crystal Violet in distilled water. Excess dye was flushed with water and then eluted for 10 min with 30% acetic acid (100 μl/well). The optical density was determined in each well by spectrometry at 540 nm. Cell survival was expressed as a percentage compared to nontreated cells.

Assessment of renal function and tissue accumulation of cisplatin in rats

Three-month-old male BD IX rats (average weight of 330 g) were subcutaneously grafted in the prethoracic area with syngeneic DHD/K12/PROb colon tumors.20 When the tumor volumes were approximately 1 cm,3 the rats (5 per group) were intraperitoneally treated with 4 mg/kg cisplatin diluted in 20 ml of an NaCl solution, which was either hypotonic (4 g/l NaCl; 131 mOsm/l), isotonic (9 g/l NaCl; 295 mOsm/l) or hypertonic (14 g/l NaCl; 446 mOsm/l). Rats were killed 72 hr after the cisplatin injection. Arterial blood was collected from the aorta to determine the plasma levels of urea and creatinine. As osmotic variations in the papilla are wider than in the renal cortex, midsagittal kidney slices were dissected to separate a cortical area from an inner medulla area (papilla). Tissue samples were immediately processed for the histologic study or frozen until assay by atomic absorption spectrometry. To determine the platinum tumor concentration, tumor and tissue fragments were weighed and digested in a microwave digester (MLS-1200 Mega, Milestone, Sorisole, Italy). After dilution in distilled water, the platinum concentrations of digested samples were assayed by AAS.

Histologic analysis of renal damage in rats

Tissue samples were fixed in a Dubosq-Brazil solution and paraffin embedded. All analyses were performed on the same paraffin blocks. Tissue sections were stained with hematoxylin-eosin-saffron (HES) and Masson trichromic techniques. Apoptotic cells were detected using the TUNEL (Terminal deoxynucleotidyl transferase-mediated dUTP Nick End Labeling) method and an immunohistochemical assay with a cleaved caspase-3 antibody (1/100; Cell Signaling, Beverly, MA). Morphologic alterations, as well as TUNEL and activated caspase-3 immunohistochemical assays, were evaluated and semiquantitatively scored by 2 independent pathologists who were unaware of the treatment group. The semiquantitative scoring system assessed 0 if no morphologic changes or labeling were noticed, 1 if changes or labeling were focal and weak, 3 if changes were diffuse and obvious at low-power magnification and 2 for intermediate damage between 1 and 3.

Determination of urine osmolarity from cisplatin-treated patients and healthy volunteers

Urine samples were collected at each urination and kept frozen at −20°C until assay. Osmolarity was determined on a cryoscopic osmometer (Fisk, Mark3 Osmometer, Norwood, MA).

Statistical analysis

All data were expressed as means. For urea nitrogen and creatinine blood concentration or platinum accumulation in renal cells statistical comparisons were made using the 1-factor ANOVA test. For platinum tissue content results, statistical analysis was performed with 2-factor factorial ANOVA, with the 2 factors being the tissue sample (kidney, liver and tumor) and the osmolarity of the injected solution (hypotonic, isotonic and hypertonic).

RESULTS

Low-medium osmolarity enhances cisplatin-induced accumulation in and damage to human renal cells in vitro

To assess the link between reduced osmolarity and cisplatin accumulation and cytotoxicity, we used human 56/10 A1 glomerular and human HK-2 tubular cells. The platinum cellular content was higher in both glomerular and tubular cells after a 3 hr incubation in a low osmolarity (150 mOsm/l) PBS compared to the isotonic (300 mOsm/l) or the hypertonic (600 mOsm/l) solutions (Fig. 1a). In the same way, platinum accumulation in both types of renal cells was higher after a 3 hr incubation in human urine of low osmolarity (Fig. 1b).

Figure 1.

Platinum accumulation in human glomerular and tubular cells according to the osmolarity of the medium. Human 56/10 A 1 glomerular cells and HK-2 tubular cells were incubated for 3 hr with 20 μg/ml cisplatin diluted in PBS (a) or urine from a healthy volunteer (b) submitted to a water restriction-dilution sequence. Clear bars correspond to the hypotonic medium (150 mOsm/l for PBS, 133 mOsm/l for urine), dark bars to the isotonic medium (300 mOsm/l for PBS, 357 mOsm/l for urine) and dashed bars to the hypertonic medium (600 mOsm/l for PBS, 605 mOsm/l for urine). Each value is the mean of 4 measurements ± SD. An asterisk indicates a statistically significant difference (p < 0.01) between the hypotonic or hypertonic medium with the isotonic medium (1-factor ANOVA test).

Cisplatin cytotoxicity on both renal cell lines was enhanced after incubation in low osmolar PBS (150 mOsm/l), whereas hypertonic PBS had only a mild protective effect on tubular cells (Fig. 2). Hyper- or hypo-osmolarity had no influence by itself on glomerular or tubular cell viability after short-term exposure.

Figure 2.

Cisplatin cytotoxicity on human renal cells, according to the osmolarity of the incubation medium. Human 56/10 A 1 glomerular cells (a) and HK-2 tubular cells (b) were incubated for 3 hr with cisplatin diluted in PBS of varying osmolarity (triangles, 150 mOsm/l; squares, 300 mOsm/l; diamonds, 600 mOsm/l). Cell survival was assessed 72 hr after treatment. Each value is the mean of 4 measurements ± SD. An asterisk indicates a statistically significant difference (p < 0.01; 1-factor ANOVA test).

Together these results established that low osmolarity increased the accumulation and cytotoxicity of cisplatin on both human tubular and glomerular cells in vitro. Iso-osmolar conditions were found to be as protective as hyperosmolar conditions.

Injection of low osmolar solutions increased cisplatin nephrotoxicity in rats

To find out if reduced urine osmolarity was a major determinant of cisplatin-induced nephrotoxicity in vivo, BD IX rats with subcutaneous tumors were treated with intraperitoneal cisplatin diluted in 20 ml of hypotonic (4 g/l, 130 mOsm/l, group I), isotonic (9 g/l, 285 mOsm/l, group II) or hypertonic (14 g/l, 445 mOsm/l, group III) NaCl solutions in distilled water. Preliminary experiments in rats had shown that urine osmolarity varied according to the osmolarity of the intraperitoneal solution (data not shown). Blood urea and creatinine levels were higher in the cisplatin-treated animals than in the control animals. However, urea and creatinine blood levels were significantly higher in animals that received cisplatin in a hypotonic solution compared to those treated with the isotonic solution (Fig. 3). The hypertonic solution had no protective effect compared to the isotonic solution.

Figure 3.

Blood urea and creatinine concentration according to the osmolarity of the solution used for the intraperitoneal injection of cisplatin in rats. Blood was collected 72 hr after the intraperitoneal injection of 4 mg/kg cisplatin diluted either in hypotonic (4 g/m NaCl; 131 mOsm/l), isotonic (9 g/l NaCl; 295 mOsm/l) or hypertonic (14 g/l NaCl; 446 mOsm/l) NaCl solution. Control rats were left untreated. Each value is the mean of 5 measurements ± SD. An asterisk indicates a statistically significant difference (p < 0.01) between the hypotonic group and the others (1-factor ANOVA test).

Functional renal alteration was correlated to the level of platinum accumulation in the renal tissues (Fig. 4). The platinum concentration was higher in the papilla than in the renal cortex for each group. Platinum accumulation was significantly higher in both the cortex and the papilla in the group that received cisplatin in the hypotonic solution. In contrast, no significant differences were measured in the platinum content of the liver and the tumor with respect to the osmolarity of the intraperitoneal solution.

Figure 4.

Platinum concentration in rat organs according to the osmolarity of the solution used for the cisplatin injection. Organs were sampled 72 hr after the same IP cisplatin treatment as in Figure 3. Hypotonic (clear bars), isotonic (dark bars) and hypertonic (dashed bars) solutions were used. Each value is the mean of 5 measurements ± SD. An asterisk indicates a statistically significant difference (p < 0.01) between the hypotonic group and both of the other groups (2-factor ANOVA test).

Pathologic features were in accordance to the extent of the acute renal failure and the kidney platinum accumulation in the animals. Rats in groups II and III developed less interstitial and tubular damage than rats in group I. The semiquantitative scoring of the areas of capillary congestion or edema, granular casts and/or cellular shrinkage, swelling and nuclear fragmentation, known to be essential in necrosis or apoptosis, are reported in Table I. A prevailing distribution of the lesions was observed in the kidney medullar zones. Apoptosis of kidney cells was confirmed by the activated caspase-3 and TUNEL assays (Fig. 5). Apoptotic bodies were seen overall in group I, predominantly in the medullar zone. Only the animals in group I showed apoptotic changes in the cortical areas, especially in the tubular structure near the glomeruli (Fig. 5d).

Table I. Semiquantitative Analysis of the Renal Morphologic Changes in Cisplatin-Treated Rats
 Group I (hypotonic)Group II (isotonic)Group III (hypertonic)Controls
Congestion    
 Vascular0000
 Glomerular++±±0
 Peritubular+±±0
Interstitial modifications    
 Leukocyte infiltration0000
 Edema0000
Tubular casts++++++0
Tubular cell damage    
 Cortical+000
 Medulla++++++0
Cleaved caspase-3 IHC  staining    
 Cortical+000
 Medulla++++++0
TUNEL assay    
 Cortical+000
 Medulla++++++0
Figure 5.

Detection of cisplatin-induced apoptosis in rat kidney. Nuclear fragmentation and apoptotic bodies were observed in HES-stained tubular cells (a), and apoptosis was confirmed by the TUNEL assay (b) and the anti-cleaved caspase-3 immunostaining (c). Apoptotic tubular cells of the cortical area, just near the glomeruli, were only seen (anti-cleaved caspase-3) in the rats that received cisplatin in the hypotonic solution (d). Magnification was 630× for (a) and (b) and 400× for (c) and (d).

Urine osmolarity in patients receiving hydration for cisplatin and in healthy volunteers

To find out if considerations on urine osmolarity were clinically relevant, urine samples were obtained from patients receiving a cisplatin-containing regimen for various malignancies. The routine hyper-hydration regimen in our institution was 2 liters of a 50 g/l glucose solution with 4 g/l NaCl and 2 g/l KCl for 3 hr before and after cisplatin. Cisplatin was given for 30 min in 500 ml of a 9 g/l NaCl solution. There were noticeable but few well-explained variations in urine osmolarity among patients and for a given patient between the first and second courses (Fig. 6). In all patients, urine osmolarity was hypotonic (inferior to 300 mOsm/l) for a moment after cisplatin administration.

Figure 6.

Urine osmolarity in patients hyperhydrated for a cisplatin treatment. Urine osmolarity was measured during the 2 first chemotherapy courses (circles, first course; squares, second course) in patients (a–e) who were hydrated before (arrow a) and after (arrow c) a 30 min IV injection (arrow b) of 60–100 mg/m2 cisplatin. Cisplatin was diluted in 500 ml of a 9 g/l NaCl solution. Hydration was 2 liters of 50 g/l glucose solution with 4 g/l NaCl and 2 g/l KCl for 3 hr before and after the cisplatin infusion.

Influence of the hydration regimen on urine osmolarity was determined in 3 healthy volunteers (Fig. 7). A 4 liter hydration of either a 50 g/l glucose solution or a 9 g/l NaCl solution was given per os for 6 hr at a 1-week interval. Furosemide (2 per os administrations of 40 mg each) was given with the NaCl solution. Hydration with the 50 g/l solution provoked a sustained drop in urine osmolarity in all volunteers (Fig. 7a). Urine was hypotonic in 1 volunteer who received the NaCl solution and normotonic or hypertonic in the others (Fig. 7b). Urine was normotonic or slightly hypotonic when furosemide was given in addition to the NaCl solution. A definite advantage of furosemide was the immediate start of diuresis contrasting to the delayed diuresis after the NaCl hydration alone (data not shown).

Figure 7.

Urine osmolarity in 3 healthy volunteers who received an oral hyperhydration. Urine osmolarity was measured during (arrow) and after a 4-liter oral hydration for 6 hr. Hydration was 50 g/l glucose solution (a), 9 g/l NaCl solution alone (b) or associated (c) with 2 oral administrations of furosemide (F).

DISCUSSION

In this article, we demonstrate that cisplatin accumulation and cytotoxicity on glomerular and tubular renal cells depend on the urine osmolarity both in vitro and in vivo in rats. No previous study has focused on the relationship between low urine osmolarity and an increase in the nephrotoxicity of cisplatin. Hyperhydration with a hypertonic solution alone or in association with mannitol was empirically found to decrease cisplatin nephrotoxicity.6, 7 High-dose cisplatin has been given safely by using a large saline hydration and 3% saline as the drug vehicle.21 Sodium chloride was demonstrated to offer protection against cisplatin nephrotoxicity in rats.22 Precise mechanisms of these nephroprotective measures were not well defined. Renal protection offered by mannitol and saline hydration was only attributed to an increase in diuresis and thus a greater drug dilution in the urine.

Within 4 to 6 hr after an intravenous administration, protein-unbound cisplatin is ultrafiltered by the glomerules and is partially reabsorbed and secreted by the epithelial cells along the tubule.23, 24, 25 In the glomerules that are in the cortex area, urine osmolarity is close to that of the plasma. Despite the fact that glomerular cells are sensitive to cisplatin cytotoxicity in vitro, these cells are protected by the relatively high local osmolarity in this renal segment. We did not observe apoptotic glomerular cells, even in animals receiving cisplatin in a hypotonic solution. In accordance with these experimental data, massive proteinuria, a sign of glomerular dysfunction, is seldom reported after cisplatin treatment in humans. We observe here, as others have previously done,26, 27 that platinum accumulation is greater in the medulla area, which contains tubular and interstitial cells, than in the renal cortex. Urine osmolarity varies largely along the tubule according to the hydration status, the action of antidiuretic hormones and the rate of sodium reabsorption.28 Osmolarity follows an increasing gradient from the cortex (300 mOsm/l) to the inner part of the medulla due to the countercurrent mechanism (until 1,200 mOsm/l). In contrast, urine osmolarity decreases to as low as 100 mOsm/l in distal convoluted tubules due to the impermeability of this segment to water and the active reabsorption of sodium by Na/K ATPase.29 Such variations in local urine osmolarity could explain the increased toxicity of cisplatin on particular tubular segments, mainly those of the medulla area where osmolarity variations are the largest. Moreover, cisplatin-induced tubular damage results in an alteration of the capacity of the kidney to concentrate urine.30, 31 The resulting decrease in urine osmolarity and free-water reabsorption could be a mechanism of the auto-aggravation of cisplatin toxicity with repeated treatment. Cisplatin nephrotoxicity is characterized by proximal and distal tubule and collecting duct dysfunction.32, 33 Cisplatin preferentially accumulates in cells of the S3 segment of the renal proximal tubule and becomes toxic intracellularly by hydration.27 Different portions of the nephron display variations in sensitivity to cisplatin, with tubular cells from the S1 segment being the most sensitive in vitro.34 The earliest manifestation of toxicity is the inhibition of protein synthesis, but many other mechanisms have been implicated, such as the activation of the mitochondrial apoptotic cascade, the generation of oxygen free radicals or the depletion of glutathione.33 Regardless of the mechanism, a resulting dose- and duration-dependent apoptotic pathway is directly involved in the pathogenesis of the cisplatin-induced renal tubular cell death process and renal dysfunction.35, 36, 37 We confirm here that cisplatin-induced renal cell damage corresponds in vivo to an apoptotic process that was characterized by caspase-3 cleavage and DNA fragmentation observed through the use of the TUNEL assay. Morphologic changes and apoptotic bodies were predominant in the medulla area and were more numerous in the group of animals that received cisplatin diluted in the hypotonic solution. The extent of the morphologic alterations was closely related to the level of tissue platinum content and the degree of renal insufficiency in this animal group.

Our in vitro experiments show that cellular accumulation depends on both the cisplatin concentration and osmolarity of the incubation medium. In the kidney, tubular cells are directly exposed to ultrafiltered cisplatin, and urine osmolarity could be a major parameter of cisplatin absorption. Only an isolated, perfused tubule technique would have given direct proof for this assumption. Increased drug accumulation in the kidneys and tubular toxicity in rats receiving cisplatin in a hypotonic solution is an indirect proof of this deleterious effect of low urine osmolarity. There is some controversy on the potential reduction of the antineoplastic activity of cisplatin administered with a high salt concentration in the vehicle.38 In several tumor systems, the survival of mice given cisplatin IV was reduced by 50% to 60% when the NaCl concentration in the vehicle was raised from 0.9% to 4%. In contrast to the kidney, we did not observe a significant difference in the tumor and liver content of platinum in cisplatin-treated rats, regardless of whether the cisplatin was given in a hypotonic, isotonic or hypertonic saline solution. A preliminary study in rats has demonstrated that plasma osmolarity decreased or increased only less than 10% compared to controls after a hypotonic or hypertonic IP load, whereas urine osmolarity varied greatly. Such a low change in osmolarity was insufficient in producing a significant modification of cisplatin accumulation in tumor cells (data not shown). In our opinion, fear of a reduction in cisplatin accumulation and toxicity in tumors should not prevent giving osmotic molecules for nephroprotection.

In conclusion, not only the urinary flow rate but also the urine osmolarity must be kept high both during the infusion and in the few hours after the cisplatin administration. We observed that urine osmolarity can be low in patients who received a routine hydration for a cisplatin treatment in our institution. The relationship between low urine osmolarity and an alteration of renal function has not been established in this short series. However, the routine hydration regimen has been modified after these data. The 50 g/l glucose solution with 4 g/l NaCl and 2 g/l KCl has been replaced by the 9 g/l NaCl solution. We now plan to check urine osmolarity after the addition of mannitol and furosemide to the NaCl hydration. Caution should be advised with the use of diuretics, noticeably furosemide, which lower urine osmolarity.23, 37 However, our preliminary results in healthy volunteers showed that furosemide could help to maintain a high urine output with sufficient osmolarity. Measurement of urine osmolarity by the cryoscopic method is an easy parameter to follow. It could be closely monitored before and during the infusion of cisplatin in order to correct any dangerous urine hypo-osmolarity by infusion of osmotically active molecules. Such a precaution could lead to a still greater reduction of histologic damage and better long-term renal function in patients cured of cisplatin-sensitive tumors.

Acknowledgements

We thank the French League Against Cancer (Committees of Côte d'Or, Nièvre and Haute-Marne). We thank Dr. F. Delarue (INSERM U 489, Paris, France) for the kind gift of the 56/10 A1 cells, M. Moutet for her technical help in histology and Mr. J. Ewing and Ms. S. Lemaire for their help in revising the manuscript.

Ancillary