pH-mediated molecular differentiation for fluorimetric quantification of chemotherapeutic drugs in human plasma

pH-mediated molecular differentiation for fluorimetric quantification of drugs in plasma. and active micromolar to At present, drug dosage is based on standardised approaches that disregard pharmakokinetic differences between patients and lead to non-optimal efficacy and unnecessary side effects. In this work, we demonstrate the potential of pH-mediated fluorescence spectroscopy for therapeutic drug monitoring in complex media. We apply this principle to the simultaneous quantification of the chemotherapeutic prodrug Irinotecan and its active metabolite SN-38 from human plasma across the clinically relevant concentration range, i.e. from micromolar to nanomolar at molar ratios of up to 30:1. (TDM), to real of in the a precise

At present, drug dosage is based on standardised approaches that disregard pharmakokinetic differences between patients and lead to non-optimal efficacy and unnecessary side effects. In this work, we demonstrate the potential of pH-mediated fluorescence spectroscopy for therapeutic drug monitoring in complex media. We apply this principle to the simultaneous quantification of the chemotherapeutic prodrug Irinotecan and its active metabolite SN-38 from human plasma across the clinically relevant concentration range, i.e. from micromolar to nanomolar at molar ratios of up to 30:1.
Therapeutic drug monitoring (TDM), i.e. the preferably close to real time measurement of medication concentration in the blood, allows to define and maintain a precise therapeutic window for drug administration, which is of particular importance for highly toxic or expensive medication used in chemotherapy and personalised medicine. 1 So far, drug dosage is typically calculated on the basis of the body surface of the patient without patient-specific analytical feedback. 2 The significant pharmacokinetic variability between patients results in either ineffective treatment or adverse toxic effects and thus failure to reach the therapeutic goal for a large percentage of drugs being administered. 3,4 The gold standard for TDM is high-performance liquid chromatography (HPLC) combined with a suitable detection method, such as mass spectrometry, spectrophotometry or spectrofluorimetry, for which many protocols are available. 5,6 Nevertheless, HPLC exhibits some inherent drawbacks, in particular the need to operate in a specialised analytical laboratory with off-site transport and long waiting times as well as elaborate sample preparation and analysis leading to overall high costs. 7 Another problem, which is often underestimated, is the variability in pre-analytic sample handling and the related inconsistencies between the measured drug concentration and the real drug concentration in blood. 8 Consequently, a simplified drug monitoring that is closer to the point of care would offer a better definition of the therapeutic window as well as continuous patient feedback for optimised efficacy. Fluorescence spectroscopy is a valuable technique used as a detection method for many chemical, biological, and medical applications. 9,10 Spectrofluorimetry offers advantages such as low detection limits, sometimes down to parts per billion or lower, as well as target specificity since simultaneous light absorption and emission at particular wavelengths provides a more discriminative route in comparison to absorption spectroscopy. Indeed, many HPLC-based protocols in TDM rely on fluorimetric quantification of the target drug. 11 Fluorimetry is also commonly used for bioimaging applications. 12 To this end, the use of pH-sensitive fluorescent probes offers a powerful route to spatially resolve the intracellular pH for the study of cell signalling and to identify diseases. 13,14 In reverse, it is reasonable to assume that a deliberate and controlled manipulation of the pH may facilitate the quantification of drugs that exhibit strong pH-dependent properties.
In this work, we present a route to make use of the distinct pH-dependent fluorescence properties of the chemotherapeutic prodrug Irinotecan and its active metabolite SN-38 to selectively quantify both compounds at clinically relevant concentrations. The method relies on analytical discrimination through a pH-induced bathochromic shift in the emission spectra. This is, to the best of our knowledge, the first example of a pH-mediated molecular differentiation for the rapid fluorimetric quantification of a mixture of two drugs in human plasma at clinically relevant concentrations. Irinotecan (7-ethyl-10-[4-(1-piperidino)-1-piperidino] carbonyloxycamptothecin, CPT-11) is commonly used as an antitumor drug, in particular for colon cancer, 15 and to a less extent for lung cancer. 16 The drug is considered a prodrug since it undergoes cleavage of the bispiperidino-side chain by carboxyesterase to form SN-38 (7-ethyl-10-hydroxycamptothecin), an active metabolite that has shown to be up to 1,000 more potent at inhibiting topoisomerase I than the parent Irinotecan. 17 Some of the main challenges for the simultaneous quantification of Irinotecan and SN-38 reside in the fact that their absorption and fluorescence properties are almost identical under physiological conditions, and the fact that the concentration of SN-38 can be over 30 times lower than that of Irinotecan. 18 The pH of the media in which molecules are dissolved can play a major role not only in properties such as their solubility, but also in their optical performance. For Irinotecan and SN-38, this was investigated by characterising the absorbance as well as the emission behaviour at excitation wavelengths of 370 nm and 430 nm under acidic (pH = 1.4) and basic conditions (pH = 12.1). As shown in Figure 1 and Table 1, Irinotecan exhibited near identical optical properties for both absorbance and fluorescence under acidic and basic conditions, while SN-38 displayed a pronounced shift for absorbance as well as emission when basifying.  A rationalisation of the molecular structure under acidic and basic conditions is depicted in Scheme 1. The difference between the chemical structure of Irinotecan and SN-38 lies in the substitution of the bis-piperidino alkyl chain present in Irinotecan for an alcohol group in SN-38. For Irinotecan, the electronic conjugation of the camptothecin core therefore remains similar in acidic and basic pH (Scheme 1, top). This is not the case for SN-38, where the aromatic alcohol is deprotonated at basic pH, forming the alkoxy ion (Scheme 1, bottom). This alkoxy ion pushes the electronic density towards the camptothecin core, thus shifting the absorbance and fluorescence spectra of SN-38 to longer wavelengths. Please see Figure S1-S3 in the Supplementary Information for 1H NMR spectra of Irinotecan and SN-38 under both acidic and basic conditions, respectively. It is worth noting that both Irinotecan and SN-38 exhibit a pH-dependent lactone/carboxylate equilibrium. 19 However, this did not affect the results presented in this work, as the lactone/carboxylate moiety is not conjugated with the camptothecin core. Under basic conditions but with a pH below 11 the aromatic alcohol group was found to be only partially deprotonated and, consequently, the pH had a very strong effect on the intensity of the fluorescence spectrum (See Figure S4, left, in the Supplementary Information). Above a pH of 11, the deprotonation of the alcohol group was almost completed which translated to a stable fluorescence spectrum at higher pH values. The fluorescence of SN-38 was only investigated at pH below 13 due to concerns regarding the stability of the molecule. At a pH of 12 the drug was found to be very stable for prolonged periods, with no appreciable signs of degradation being observed within 24 hours. Control over the pH, in particular a reliable basification, was of crucial importance for this work. A buffer was prepared by dissolving 98 mg of NaOH and 372 mg of KCl in 100 ml of MeOH, resulting in a pH value of 12.3 as determined by a pH meter with aqueous calibration. This buffer will be referred to hereafter as PSB (potassium chloride and sodium hydroxide buffer). The ability of PSB to maintain a constant basic pH was verified by adding different volumes of the buffer to a solution of 0.05 % (v/v) trifluoroacetic acid in MeOH (hereafter TFAM, pH = 1.4). As an example, a 1 to 1 (v/v) mixture of TFAM and PSB (V t = 8 ml) resulted in a pH value of 12.1, with volumetric variations of ± 1 ml in the amount of PSB having virtually no effect on the pH of the media (See Figure S4 in the Supplementary Information)). The viability of using a pH-mediated fluorimetric quantification of both compounds was further investigated by measuring the fluorescence of Irinotecan and SN-38 in MeOH over a range of relevant concentrations. Values previously reported in patients ranged from 10 to 10000 ng/ml (17 to 17000 nM) for Irinotecan and 1 to 500 ng/ml (2.5 to 1270 nM) for SN-38. 20 Consequently, a range from 2 to 2500 nM was studied for Irinotecan and 0.5 to 300 nM for SN-38. While the fluorescence of Irinotecan was measured in acidic conditions (pH = 1.4) using an excitation wavelength of 370 nm, the fluorescence of SN-38 was determined in basic conditions (pH = 12.1) with an excitation wavelength of 430 nm. As shown in the Supplementary Information (Figure S5), the fluorescence intensity of both compounds was found to be linear (R 2 = 0.999 for both Irinotecan and SN-38) across the range of investigated concentrations. Thus, the observed sensitivity for fluorimetric detection greatly exceeded previously reported requirements. 20 Solid phase extraction (SPE) is a viable alternative to conventional sample treatment protocols based on liquidliquid extraction, offering benefits such as faster, less labourintensive sample processing, reduced solvent use and higher concentration factors. 21 Supplementary Information, Figure S5). The sample processing translated to an overall dilution of the concentration of Irinotecan by a factor of 6.9 x in the aliquot where fluorimetric quantification was carried out. The dilution factors for the individual processing steps were 2 x for the SPE dilution, 2 x for the dilution with TFAM, and 1.7 x to account for the Irinotecan recovery efficiency during SPE. This compared to a dilution factor of 8 x for SN-38 when quantified under basic conditions, which was based on individual factors of 2 x for the SPE dilution, 2 x for the dilution with TFAM and a final 2 x for the basification with PSB. Due to the close-to-unity extraction efficiency, no SPE loss factor was necessary for SN-38. Alongside, the entire protocol was carried out with a reference sample that contained only pooled plasma in order to account for the parasitic fluorescence of the plasma ( Figure  S6, see Supplementary Information).
The results are summarised in Figure 2. In total seven samples were run in triplicate, six spiked and one reference sample. For Irinotecan, 17 out of 18 obtained data points were within 15% error (94%). In the case of SN-38, 16 out of 18 data points were within this tolerance (89%). An overview of the fluorimetric signal intensity and errors obtained across the range of samples can be found in the Supplementary Information (Tables S1 and S2). We note that for samples with concentrations below 100 nM for Irinotecan and 15 nM for SN-38, the measured background signal of the reference sample became significant, resulting in higher errors below this level, which is at the bottom end of the relevant clinical range.
We want to point out that our method is based on fluorimetric quantification and thus relies on an adequate manipulability of the emission properties of target compounds. In order to show the applicability to other commonly used chemotherapeutic drugs with similar molecular structure, we compared the pHdependent optical properties of SN-38 to Epirubicin and Methotrexate. [26][27][28] With the former containing an alcohol and the latter an amine group attached to the aromatic core, a basic pH is likely to cause deprotonation resulting in a change of the optoelectronic properties of the molecules. The absorbance and fluorescence spectra of the three compounds were recorded in acidic (pH = 1.4) and basic (pH = 12.1) conditions (Table S3, Figure S10,

Conflicts of interest
There are no conflicts to declare.

Optical measurements:
The absorbance spectroscopy set-up included a laser-driven light source (Energetiq Technology, EQ-99XFC LDLS), a temperature-controlled cuvette holder (QNW, Q-Pod) and a high sensitivity spectrometer (Ocean Optics, QEPro) as the detector. Fibre optic cables with a core diameter of 1000 µm (Ocean Optics) were used to transmit the light through the system. Disposable

ml cuvettes made of polystyrene served for absorbance measurements (Fisher Scientific).
Fluorescence measurements were carried out with a spectrofluorimeter (Shimadzu, RF-6000) using a slit width of 5nm and 10 x 10 mm quartz cuvettes (Hellma, Suprasil).

Solid phase extraction:
Solid phase extraction experiments were carried out using disposable extraction columns (Bakerbond, SPE octadecyl (C18)) with a size of 3 ml and a sorbent weight of 200 mg, purchased from VWR. The SPE protocol was carried out as follows. First, three volumes of 1.5 ml (1 st step TFAM; 2 nd step TFAM/TFAH (1/1), 3 rd step TFAH) were passed through to condition the column. Then, 0.5 ml of spiked plasma was loaded into the column. The drugs would be previously dissolved in MeOH and spiked in such a way that the amount of MeOH in plasma did not exceed 5% by volume to avoid denaturisation of the plasma content. Subsequently, three volumes of 1.5 ml (1 st step TFAH, 2 nd step TFAM/TFAH (1/9), 3 rd step TFAH) were run to wash the column from its plasma content. Finally, 1 ml of TFAM was run through the column to elute the drugs. The conditioning step included an initial wash with MeOH in order to wet the surface of the sorbent (C18, endcapped) and to penetrate bonded alkyl phases, which then allowed the water to wet the silica surface efficiently. After the conditioning step, the plasma with the spiked drugs of desired concentration was loaded into the cartridge. The three aqueous washing steps were introduced to remove the vast majority of the plasma content and thus avoid parasitic fluorescence when eluting the drugs. The second wash contained 10 % of MeOH by volume in order to promote the elution of plasma content that was not soluble in pure water. This percentage was found to be optimal as a smaller content of MeOH resulted in a stronger parasitic signal from the plasma, and a higher content of MeOH led to unreliable drug recoveries, most likely due to partial elution of the drugs.

Quantification of Irinotecan and SN-38:
Irinotecan + SN-38: After dilution of the SPE extract with 1 ml of TFAM, the fluorescence was measured at 432 nm with an excitation wavelength of 370 nm.
SN-38: Subsequent to measuring the fluorescence of the drugs in acidic conditions, the sample was basified by adding an equal volume of PSB. The fluorescence was then measured at 559nm with an excitation wavelength of 430nm. This measurement was compared with the calibration curve made for SN-38 taking in consideration the dilution factor of 8 x to derive the concentration of SN-38.
Irinotecan: First, the established amount of SN-38 was translated into a corresponding emission in acidic conditions based on a previously built calibration curve and subtracted from the overall intensity. The corrected value was then compared with the calibration curve made for Irinotecan taking in consideration the dilution factor of 6.9 x to obtain the concentration of Irinotecan.
NMR Spectroscopy: NMR Spectroscopy was carried out at 297.2K on a Bruker Avance III 500 spectrometer at frequencies of 500MHz for 1H nuclei.           Table S4. Chemical structure and maximum intensities for the absorbance and fluorescence in acidic and basic conditions of SN-38, Epirubicin and Methotrexate. The maximum in absorbance was used as the excitation wavelength for the fluorescence spectra.