A Model of Chemotherapy-Induced Tumor Dormancy: Doxorubicin Retention and Degradation Denes Resumed Growth in Vivo

Background. Cancer recurrence after adjuvant chemotherapy with long periods of remission is common. After cessation of the therapy, the dormant cells may repopulate, but the signals that control the tumor to exit dormancy are not completely understood. We hypothesized that tissue bound cytotoxic drug indwelling in a body for years might somehow contribute to dormancy or recurrence of a tumor. The purpose of this study was to design a model demonstrating that viable cells implanted in mice can repopulate or be suppressed, depending on the cytotoxic preload conditions. Methods. A two-step dormancy/recurrence (TSDR) murine model was designed, which mimics the extrusion of taken up drug from tumor cells. The viable cells preloaded with drug were implanted into mice. The survival rates of these mice were then used as criteria to demonstrate the relationship between resumed growth and drug cellular eux/viability. The drug internalization patterns following their exposure to doxorubicin (dox) or degraded dox (dox-dgr) were investigated by exploring ow cytometry, spectral analysis, high performance liquid chromatography and confocal microscopy. Antiproliferative and myelotoxic capacity was evaluated by hematological nadir induced by the iv injected drug. Results. The viable SL2 lymphoma cells exposed to 10 µg/ml of dox for 30 min and injected into syngeneic DBA/2 mice were made unable to recure. Exposure of cells to lower dox concentrations (0.01 – 1.0 µg/ml) resulted in tumor recurrence, similar to that which was observed during implantation of an untreated tumor. Dox-dgr kept at 37 °C for 365 days lost its tumoricidal and antiproliferative capacity and displayed a loss of selectivity of nuclear uorescence. nude mice


Background
Residual tumor cells are detected in most cancer patients following surgical and adjuvant treatment (1).
These residual cells might remain dormant for years before resuming proliferation and recurrence -the leading cause of cancer mortality (2,3). A little known fact is that several anticancer drugs might also remain in the body for many years following successful chemotherapy. The most studied agents in this context are platinum compounds continuing to be present in blood and urine of patients treated more than 20 years ago (4). The long duration of tissue bound platinum was hypothesized to be related to late neurotoxicity, increased secondary cancer risk, tinnitus and some other side effects (4,5). We hypothesized that prolonged presence of tissue bound cytotoxic drug might be associated with not only side effects, but also with dormancy prolongation. Interestingly, platinum excreted into plasma of patients was found to be pharmacologically active and retained some 10% of initial reactivity in ex vivo assays up to 3 years of body indwelling (6). Some indirect evidence of antitumor activity of tissue retained platinum is found in surgical specimens taken from patients who received preoperative chemotherapy. Patients with higher platinum concentration in a lung tumor had a longer time until recurrence and longer overall survival (7). Platinum was detectable in surgical specimens up to 72 days after preoperative chemotherapy and higher platinum content in tumors correlated with improved pathological response of gastric cancer (8). In fact, long term tissue presence was found in autopsy specimens of patients treated not only with platinum compounds. Different groups of cytotoxic drugs or their metabolites were also shown to exhibit long tissue persistence in autopsy specimens (9,10,11).
However, all of these studies were focusing primarily on the drug elicited side effects.
The majority of cytotoxic drugs are known to exhibit a high volume of distribution, short half-life in plasma and high tissue uptake following a single iv shot. Cancer chemotherapy obviously spares some malignant cells and the outcome of micrometastatic disease might be related to miniscule amounts of pharmacologically active drug located in tissues and cells. This tissue retained drug is exposed to multiple factors devaluing its antitumor activity: enzymes, pH uctuations, reactive radicals, body temperature, etc. The complexity of these mechanisms and long duration of factor interplay is not easy to reproduce in an experiment and the subject has not yet been studied.
We developed an assay to demonstrate the role of cellular drug internalization on tumor recurrence rates.
Viable SL2 lymphoma cells loaded with doxorubicin (dox) were exhibiting implantation and growth when implanted intraperitoneally (ip) into DBA/2 mice if the exposure dose was lower than 10 µg/ml. Long term body temperature degraded dox lost its antiproliferative capacity, indicating that tissue retained drug might not be effective enough to assure tumor dormancy.

Materials And Methods
Mice, tumors, cell line, drugs testing DBA/2 mice, aged 8 to 12 weeks, were obtained from a local breeding facility SRICIM, Vilnius, Lithuania. Mice were used and cared for in accordance with the Guide for the Care and Use of Laboratory Animals.
All research protocols were approved by the Institutional Animal Care Committee. Animals had ad libitum access to pelleted feed, food supplements and water. All animals were determined speci c pathogen free.
Dox or dox-dgr were injected iv via retroorbital plexus as a single push at a dose of 15 mg/kg. Light inhalation anesthesia with iso urane was used for iv injections and blood sampling procedures. The SL2 lymphoma occurred as a spontaneous tumor in a DBA/2 mouse of the Chester Beatty Research Institute, UK, and was acquired through prof. Den Otter, University Medical Center Utrecht, Netherlands. SL2 cells were maintained in RPMI-1640 medium (Thermo Fisher scienti c, Waltham, USA) containing 10% heatinactivated fetal bovine serum (Thermo Fisher scienti c, Waltham, USA). All tumors in our TSDR study were induced by ip injection of 5x10 5 cells per mouse. The therapeutic e cacy of dox and dox-dgr was tested by injecting drug iv at speci c time intervals (24 hr, 48 hr, 120 hr, 168 hr and 192 hr) following ip implantation of 5x10 4 cells on day 0.

Drugs and reagents
Doxorubicin hydrochloride (2mg/ml) (Ebewe, Unterach, Austria) was obtained from a hospital pharmacy. Daunorubicin hydrochloride (analytical grade), chloroform (HPLC grade). Acetonitrile, orthophosphoric acid and hydrochloric acid were purchased from Sigma Chemical Company (St Louis, MO, USA) and HPLC grade methanol was purchased from Carl Roth GmbH (Karlsruhe, Germany).
Two step tumor dormancy/recurrence (TSDR) model The time course of dox-loaded viable SL2 cell ip implantation into DBA2 mice is shown in Fig 1. SL2 cells were cultured for 48 h to obtain 5x10 7 cells (viability > 95%). Then the suspension was washed and incubated for 30 min. within RPMI-1640 medium containing 10% of fetal bovine serum (FBS) and variable amounts of dox or dox-dgr (10 µg/ml, 1.0 µg/ml, 0.5 µg/ml, 0.1 µg/ml, 0.05 µg/ml). Following 30 min. of incubation, the media containing dox or dox-dgr was washed with phosphate buffered saline (PBS) and put on ice before the transplantation procedure. Then SL2 dox loaded cells were diluted in PBS and a total of 5x10 5 cells per mouse (inoculation volume 0.2 ml, viability > 95%) were injected ip. The viable cell number was determined by performing trypan blue stains on the cells. Control mice received the same number of cells exposed to RPMI-1640 medium containing no dox. Total time of the manipulation procedure from cell harvesting to inoculation into mice did not exceed 60 min. The number of surviving mice in each experimental group was checked daily until day 30, then checked twice a week until day 60. All mice that died had an evident ascitic tumor. Mice surviving more than 60 days were considered as cured.  was used as an uptake value. Non-viable cells and duplets were excluded by gating out their light scatter characteristics. A total of 20,000 events to generate each histogram were acquired.

Confocal microscopy and spectral analysis
Following exposure to dox and dox-dgr, SL2 cells suspended in PBS were made adherent to glass coverslips placed at the bottom of a 12-well culture plate (TPP AG. Trasadingen, Switzerland) via 30 min. of gravity sedimentation as described by Tsang et al. (12). Adhered cells were xed in 4% paraformaldehyde and then analyzed. The localization of dox and dox-dgr in SL2 cells were analyzed under a confocal microscope (Nikon Eclipse TE2000-S, C1 plus, Nikon, Tokyo, Japan) by scanning with the argon ion laser (488 nm) using oil immersion 60× NA 1.4 objective (Plan Apo VC, Nikon, Japan). Bright eld microscopy was performed to visualize morphology of the cells. The uorescence of dox and dox-dgr was detected using a 515/30 bandpass lter (Semrock Inc., USA). In addition, spectrally-resolved intracellular uorescence of dox and dox-dgr was registered using a microscope 32-channel spectral imaging unit. The uorescence spectra of dox and dox-dgr in single cells and subcellular regions was analyzed. In the region of interest (ROI) in the stacked confocal images, cell uorescence was measured. The images were further processed using EZ-C1 v3.91 (Nikon, Japan) and ImageJ v1.53a software (NIH, USA).

Optical characterization
Steady-state absorption and uorescence spectra of doxorubicin in PBS were measured with a singlebeam spectrophotometer Varian Cary Win UV (Varian Inc., Australia) and spectro uorometer (FLS920, Edinburgh Instruments, Livingston, UK). Polystyrene cuvettes with an optical path length of 1 cm were used for all measurements.
Blood sampling, plasma preparation, complete blood counts Blood samples (125 µL) were taken from the retroorbital plexus before and after the dox administration at several time points 5 min, 15 min, 30 min, 60 min, 6 hr, 12 hr, 48 hr and 72 hr in tubes containing EDTA as an anticoagulant. Dox-dgr measurements were not performed due to imprecision of the procedures (presence of byproducts with HPLC retention times close to dox and overlapping with internal standard). Blood sampling and dox injection were performed on two contralateral retroorbital plexus, while iso urane anesthesia was applied. Each blood sample was gently inverted several times to ensure complete mixing with the anticoagulant. After centrifugation at 5000 x g for 10 min, plasma samples were separated and stored at -20 °C until analysis. Complete blood counts (CBC) were analyzed using ABX Micros ESV 60 within an hour of sampling.

Pharmacokinetics
The microvolume method (13) was applied in this study for plasma Dox analysis. 50 µL of internal standard solution (800 ng/mL Daunorubicin hydrochloride) was added to 60 µL of a plasma sample and vortex-mixed for 30 seconds. The extraction of the drug was performed by adding 900 µL of a chloroform/methanol mixture (4:1, v/v). After vortex mixing for 10 min and centrifugation (10 min, 10,000 x g), the organic phase was collected, transferred to a clean tube, then evaporated to dryness under a stream of nitrogen. Dry residue from the plasma was dissolved in 60 µL of mobile phase, and after centrifugation for 5 min (10,000 x g), injected into the chromatographic column. The area under the concentration curve (AUC) were calculated by trapezoid rule.

Chromatographic conditions
High performance liquid chromatography (HPLC) analysis was performed on a Perkin Elmer system that consisted of a Flexar binary LC pump, a Kit -Flexar 3 CHNL VAC degasser, a Flexar LC autosampler and Flexar uorescence detector (Xenon lamp). The chromatography data was acquired by Chromera software from Perkin Elmer. The chromatography separation was performed on a Brownlee Bio C18 column (4.6 × 150 mm, 5 μm particle size, Perkin Elmer, Shelton, USA). The optimum mobile phase consisting of acetonitrile and water (32:68, v/v), was pH adjusted to 2.6 with 85% orthophosphoric acid. Samples were delivered via isocratic ow at a rate of 0.25 mL/min. The column temperature was maintained at 37 °C and excitation and emission wavelengths were set at 475 and 555 nm respectively. The injection volume was 50 µL.

Statistical analysis
The statistical tests were performed using STATISTICA 12.0 (TIBCO Software Inc Palo Alto, California, USA).
All the results are presented as means and standard error (mean ± S.E.). Signi cance was considered at values of p < 0.05.

Results
The characteristics of the SL2 cell line used in our TSDR model are presented in Table 1. We have chosen to use SL2 lymphoma line syngeneic to DBA2 mice to investigate the relationship between dox uptake and tumor dormancy control in vivo. The rationale for choosing SL2 lymphoma was the following: 1) sensitivity to dox in vivo and in vitro, 2) exact range of dox vs tumor size shows therapeutic effectiveness i.e. cure (small tumors) and recurrence (medium or large tumors), 3) speci c immunophenotype allowing minimal residual disease monitoring in ascite samples, 4) suspension type of growth in vitro, allowing to quantitatively analyze dox uptake by FCM without detachment, trypsinization or additional manipulation. This is of particular importance when exploring our TSDR model and attempting to shorten manipulation time. Short high dose exposure helps to keep high and comparable cell viability values before and after dox loading. In addition, adherent subtypes of identical cell lines growing in suspension might exhibit signi cant variability of dox resistance (14). The SL2 lymphoma belongs to the T cell lineage and was reported to express Thy 1.2 (15). The SL2 line explored in our TSDR model maintained its T cell speci city with CD3+4-8+44+45+ immunophenotype. This phenotype alongside FCM light scatter characteristics, permits tumor cell distinction in ascites samples with high host cell admixtures.
Chemotherapy simulation in vitro is usually experimented with drug exposure for several days (16,17). We preferred to simulate the short initial (distribution) pharmacokinetic phase since the uptake in vivo is mostly related to it. The second phase of biexponential elimination curve is more attributable to a dox tissue e ux rather than the uptake. We attempted to modify already reported dormancy/recurrence model (16) by signi cantly shortening the exposure duration and by implanting cells into mice rather than returning them to a cell culture (schematic of our TSDR model shown in Fig 1). The lowest effective dox concentration continuously imitating the dormancy in vivo was 10 µg/ml (Fig 1). To further investigate how this concentration compares to plasma dox content, we performed pharmacokinetic analysis of dox injected iv Unexpectedly, maximal plasma dox concentration was never reaching the 10 µg/ml (Fig 2) needed for e cient in vivo tumor dormancy effect (C max was 1.33 µg/ml). AUC 0-0.5 during in vitro dox loading was 5.0 (µg•h/ml) whereas plasma AUC 0-72 of iv injected dox was 3.6 (µg/ml • h). Failure of doxdgr to affect any hematopoietic cell was observed comparing HGB, WBC and PLT counts in all three animal groups (control, receiving dox or dox-dgr). Dox induced a dramatic leukopenia as WBC counts decreased signi cantly from 7.7+0.59 k/µl to 2.6+0.42 k/µl. Dox-dgr did not retain the capability to induce leukopenia (WBC counts were 7.1 + 1.4 k/µl, p=0.4).
Tissue retention of cytotoxic drug is known to persist for several years. To simulate long term tissue indwelling of dox, we incubated the drug at 37 °C in darkness in the original vial (sterile) for a period of 365 days. Body temperature induced drug decay was analyzed by HPLC and by spectrometry following 120 days (medium term) and 365 days (long term) exposure at 37 °C. The characteristics of dox-dgr are summarized and compared to fresh dox in Table 2. The broadening of the main absorption band with signi cant hypochromism in the spectral region between 415 nm and 540 nm, and a weak hyperchromism in the spectral region above 540 nm, is an absorbance modi cation of dox-dgr which occurred during storage for the medium-and long-term periods at 37 °C (spectrum presented in supplemental material). The observed absorbance modi cation and signi cant loss of uorescence intensity in aqueous solutions resembles characteristics of aggregation and/or dimerization of dox (18). The uorescence spectra of dox and dox-dgr were also compared in samples of the aqueous solution and in the nucleus of drug loaded SL2 cells. Presence of nuclear dox or dox-dgr determined by confocal ROI analysis exhibited uorescence with spectral characteristics, de ned by three distinct peaks similar to the aqueous solution spectra of both drugs (Fig 3, A and B). However, the dramatic quantitative reduction of dox related uorescence was seen in the aqueous solution of dox-dgr, when compared to dox samples (Fig 3 C). The HPLC analysis of the aqueous solutions of dox-dgr revealed that the characteristic dox peak is signi cantly reduced compared to dox sample (Fig. 3 D), but still noticeable and exhibits a retention time similar to some byproducts. (Fig 3 E and F). The uptake of dox-dgr resulted in dim but still detectable uorescence in SL2 cells after in vitro drug load before proceeding with second in vivo step of TSDR procedure (Fig 4 a-c compared to Fig 4 d-f). The nuclear uorescence intensity measured by confocal microscopy in regions of interest (ROI) were signi cantly lower in dox-dgr samples when compared to fresh dox samples. Notably, dead/apoptotic cell presence showed hyperchromatism of nuclear uorescence due to the dox-dgr (Fig 4 g -i), which is known to be a feature of fresh dox in permeable cells. The dim dox-dgr uorescence lost its nuclear selectivity in viable cells and was found to be homogenously distributed between their cytoplasm and nucleus. The dox-dgr also lost its capability of inducing a myelotoxic (antiproliferative) effect after iv injection compared to dox (Fig 2).

Discussion
In this study we used dox, a mainstream chemotherapeutic agent, which is inherently uorescent, as a model drug to demonstrate its cellular uptake signi cance in our TSDR model. Dox is known to enter a cell, intercalate to DNA and block topoisomerase II. The cell's ability to proliferate is then strictly conserved to a number of intracellular dox molecules attained by DNA (19). The question is if this intracellular drug indwelling in tissues for years (and recycling locally) can keep controlling a minimal residual disease. For more than ve decades, dox is widely used as a front-line therapy for several different malignancies. However, its role in controlling residual micrometastatic diseases (if any) is not completely understood. Chemotherapy does not only spare some viable tumor cells but also results in a long tern tissue persistence of injected drug. The cellular bound drug and its active metabolites de nitely undergo some gradual transformation while indwelling in tissues. The characteristic feature of this transformation and its effect on tumor and host tissue is not studied. It can feature decreased or increased antitumor (and dormancy maintaining) activity. Our dox-dgr samples showed obvious visual change but we analyzed only some of its physicochemical properties. Spontaneous aggregation of dox is known to induce a decrease of the uorescence quantum yield attributed to the formation of nonuorescent dimers (18) Self-association is a process that competes with binding to DNA and formation of hetero complexes (18,27).
Tumor cells loaded with dox following exposure to 10 µg/ml, lost their ability to resume proliferation in DBA/2 mice when injected IP. Exposure of cells to lower doses of dox did not result in any recurrence delay. The SL2 cell viability (> 90 %) after dox exposure of all samples before their ip implantation in mice was su cient to produce tumors. As little as 100 cells were reported to be enough to result in ascitic SL2 lymphoma growth and kill the mice (15). The rate of proliferation and aggressiveness of SL2 line is notable: the growth fraction of an SL2 tumor on day 4 after implantation is 79.2% (20). So just one viable cell out of each 5000 cells implanted (0.02% viability) should have been su cient enough to assure tumor formation and growth. We hypothesized that gradual extrusion of dox from the tumor cells should have resulted in the tumor escaping from the dormancy state attained after the initial drug cellular accumulation. A tumor exposed to 10 µg/ml dox for 30 min, was empirically determined to be su cient enough to stop tumor proliferation. Subsequently, a pharmacokinetic study was conducted to clarify how this exposure (in terms of dox concentration and AUC) compares to plasma dox concentration detectable after a 15 mg/kg dox injection. The highest plasma dox concentration was reached at 5 minutes post injection, and was only 1.33 µg/ml. It might be questionable if our TSDR model is relevant to an in vivo animal or clinical conditions. That is because we failed to demonstrate that maximum plasma dox concentration ever reached 10 µg/ml in mice following a high dose of dox administration. However, several facts are to be taken into consideration: 1) tissue and cells in mice following the injection were exposed to detectable dox plasma levels considerably longer than 30 min; dox in the plasma of treated DBA/2 mice was detectable for at least 72 hr, whereas in vitro exposure conditions lasted just 30 min, 2) the AUC 0-0.5 of TSDR model exposure was 5.0 (µg•h/ml) whereas AUC 0-72 of dox in mice plasma following a 15 mg/kg iv. injection was 3.6 (µg•h/ml); the same exposure dose (10 µg/ml) can be compared to dox uptake investigations done by others; For instance the duration of 10 µg/ml exposure of 5 days was reported (17); 4) apparently, the short tissue exposure to dox is more relevant to clinical situations than AUC (21), 5) the su cient cellular uptake of dox is already seen following 5 sec of exposure and might play a role in tumor growth control (22). It is technically impossible to test if, at a 5 sec time interval, the plasma levels are or are not reaching 10 µg/ml following dox iv administration. The max plasma concentration in humans was found to be in the vicinity of 1 -4 µg/ml after the administration of common therapeutic doses (23). However, major exposure of cells and tissue to dox in humans is found in the terminal elimination phase where plasma concentrations are around 0.05 µg/ml and continues to persist for at least 48 hr (25). Not only dox, but also its active metabolites, are found in signi cant amounts in human plasma and are contributing to its antiproliferative effects.
The take up of dox as well as a majority of other cytotoxic drugs tend to be retained in tissues for months and years. The accidentally extravasated dox was found to be present in signi cant amounts in tissue adjacent to the extravasation site for 5 months (25). Tissue uptake of dox is hundreds of times higher than the maximum plasma concentration (26). We expected that dox-dgr can partly simulate the tissue indwelling of dox and should have retained some antiproliferative capacity. The short term (21 day) dox 37 °C exposure did not reveal any absorbance or uorescence spectrum alteration in an aqueous solution or cellular nuclei (data not shown). However, at medium term (120 days), the exposure to dox-dgr lost its antiproliferative capacity and was not capable of inhibiting tumor cell recurrence in our TSDR model. Upon HPLC analysis, the residual dox peak was still noticeable, although it exhibited signi cant quantitative decline. The SL2 cells prepared for IP injection revealed identical intracellular uorescence spectra for dox and dox-dgr as judged by confocal microscopy. Fluorescence intensity and uptake patterns however, were dramatically altered in dox-dgr exposed samples as compared to fresh dox samples. There was no preferential nuclear accumulation in samples exposed to dox-dgr, which was a characteristic feature of fresh dox uptake. This raises the question if nuclear dox binding capacity is not undergoing transformation following long term body indwelling of a drug. In fact, cellular binding patterns might have a direct link to dox's therapeutic e cacy. For instance, dox cytotoxicity in a clonogenicity assay was shown to be highly dependent on cellular accumulation patterns (28). Internalized tightly bound cellular dox was the only active form of drug eliciting cytotoxic effects, as reported in this study. The tightly bound dox was not very prone to outward e ux. However, 90 % of intracellular dox could be attributed to a loose bound category demonstrated in this investigation.
Dox is known to enter the cell by passive diffusion, but at lower concentrations much of it is taken up rapidly by high-a nity binding sites (29). At higher concentrations, these binding sites are saturated and additional drug taken up by the cell remains unbound in the intracellular space. It is not clear if this intracellular unbound dox can play the role of lose bound fraction and be prone to a rapid extrusion upon IP implantation in our TSDR model. Our ndings might be questioning the supposition that long term control of micrometastatic disease is elicited through direct antitumor effects of cellular bound drug. Our assumption is based upon the fact that long term simulation of indwelling drug failed to demonstrate residual antiproliferative activity. However, this does not rule out the possibility that tissue bound weakened dox is still involved in suppressing tumor regrowth at least for several months following discontinuation of chemotherapy. This temporal effectiveness of dormancy control may alter the tumor microenvironment and activate some other host mediated tumor control mechanisms (i.e. immune surveillance). Other cytotoxic drugs, such as platinum, were shown to retain antiproliferative capacity for up to 3 years following iv therapy (6). Our ndings do not rule out the possibility that some marginal antiproliferative capacity of dox-dgr was still preserved. The WBC nadir assay of dox (15 mg/kg, iv) revealed a dramatically higher antiproliferative capacity for fresh dox compared to the dox-dgr. However, the relative weakness of dox-dgr in this assay might simply be due to a smaller amount of active dox fraction within a whole dox-dgr vial. The HPLC analysis demonstrated that dox-dgr retains only 16.8 % of the initial dox amount after 120 days at 37 °C. This means the 15 mg/kg of dox-dgr would correspond to roughly 2.5 mg/kg of fresh dox. Interestingly, 2.5 mg/ml of fresh dox injected iv into DBA2 mice did not show signi cant WBC decline at a 48 hour time point (data not shown). Even fresh dox tends to form pharmacologically inactive dimers in a vial at room temperature. Some 47 % of the fresh dox can be attributable to these spontaneously forming byproducts (18,25). We cannot rule out the possibility that DNA bound dox might retain even more antiproliferative capacity than the dox kept at body temperature in the original vial. In theory DNA intercalated dox molecules should not be forming inactive aggregates in vivo.
Dox was rather extensively studied in the models of forced degradation under the conditions of hydrolysis (water, acid and alkali), peroxide oxidation, dry heat (short duration at 80 °C), photolysis and microwave irradiation (29,30). To our knowledge, our study is the rst to report on the degradation of dox stored at body temperature for the duration of 365 days. We were not speci cally identifying the byproducts related to the dox decay, we instead tested the antitumor and myelotoxic properties of a whole dox-dgr product.
The dormancy/recurrence model developed by others with SU159 cells (16) explored a lower dox concentration (0.5 µg/ml) than ours. However, substantially longer exposure time (96 hr) was used to demonstrate dormancy in vitro. This can be converted into AUC 0-96 48 µg/ml which is signi cantly higher than the one used in our DTSR model. Interestingly this duration of exposure of dox with SU159 cells was considered to be short. Notably, the maximum plasma concentration of dox results in a peak concentration higher than 1.0 µg/ml in humans (23). In the model proposed by El-Kareh et al. (21), the dox mediated cell kill was demonstrated to be dependent on the peak concentrations rather than the time integral of concentration. We hypothesized that 96 hr in vitro exposure (16) might not entirely duplicate the in vivo situation where dox plasma levels are barely detectable. The in vitro model is for cells in culture, where oxygenation, pH, cell density, extracellular dox clearance, drug metabolism in liver and elimination in bile or urine are generally different than conditions in vivo. An empirically determined short exposure procedure which was blocking successful tumor implantation can help to better understand the mechanisms of local cytotoxic drug effects. For instance, dox at a dose of 1000 µg/ml for 60 -120 min is used by intravesical instillation for the treatment of super cial bladder carcinoma. This is a 100 times higher concentration than explored in our study. In addition, the intravesical instillation is sometimes repeated with an interval of 1 week to 1 month, depending on whether the treatment is therapeutic or prophylactic. A local injection at the maximum safe dose of dox (1000 µg in the upper lid and 1500 µg in the lower lid) was used for the treatment of blepharospasm (31). The cumulative dose in this study was even higher -4000 µg of dox per lid. Local implantable surgical devices, such as biodegradable mesh, are occasionally patented to claim therapeutic e cacy of anticancer drugs including dox (34). A dose of 10 µg/ml for 30 min de nitely remains in the safe zone for a vesicant drug such as dox when applied locally. The effective concentration of dox or any other cytotoxic drug can be additionally investigated with the help of our TSDR model. Dox controlled release via biodegradable ureteral stent (35) is an example where this model could be explored for a rapid pre-screening to determine an effective concentration range for local use.
The idea that long term tumor exposure to suboptimal doses is breeding genetically resistant cells is just one of the popular premises. This assumption is based mostly on experiments in vitro. It is a well-known fact that resistant lines can be continuously passed in media containing a cytotoxic drug and losing no proliferative capacity. The resistance in humans is not necessarily mimicking this effect. Drug intracellular presence is the most obvious feature currently known to restrain cells from resumed proliferation. Notably, the primary resistance in humans and laboratory animals might be explained for instance by inoculum effect (36) or abundance of tumor cells (37). Multiple other non-genomic resistance factors were described: non-genetic plasticity (38), drug gradients in relation to blood vessels (39), regions of hypoxia (40), speci cities of an extracellular matrix (41), abnormalities of tumor vascularization (42), and an abundance of cellular debris (43). All of these conditions are not easy to replicate in models relevant to clinical conditions. Our TSDR model is an attempt to elucidate one of the aspects of primary resistance and dormancy, which is the link between tissue retained drug and tumor recurrence.

Conclusion
The TSDR model is a rapid convenient tool to study in vivo behavior of cells preloaded with cytotoxic drug. This approach focuses on mechanisms of tumor cells exiting dormancy, in relation to cytotoxic drug e ux. The one-year body temperature exposed doxorubicin did not retain su cient cytotoxic capacity as compared to fresh doxorubicin. Exploration of in vivo model for drug extrusion analysis reveals additional potentials to study drug tissue indwelling role in dormancy.