Detection of sub-nmol amounts of the antiviral drug favipiravir in 19F MRI using photo-chemically induced dynamic nuclear polarization

In biological tissues, 19F magnetic resonance (MR) enables the non-invasive, background-free detection of 19F-containing biomarkers. However, the signal-to-noise ratio (SNR) is usually low because biomarkers are typically present at low concentrations. Measurements at low magnetic fields further reduce the SNR. In a proof-of-principal study we applied LED-based photo-chemically induced dynamic nuclear polarization (photo-CIDNP) to amplify the 19F signal at 0.6 T. For the first time, 19F MR imaging (MRI) and spectroscopy (MRS) of a fully biocompatible model system containing the antiviral drug favipiravir has been successfully performed. This fluorinated drug has been used to treat Ebola and COVID-19. Since the partially cyclic reaction scheme for photo-CIDNP allows for multiple data acquisitions, averaging further improved the SNR. The mean signal gain factor for 19F has been estimated to be in the order of 103. An in-plane resolution of 0.39 × 0.39 mm2 enabled the analysis of spatially varying degrees of hyperpolarization. The minimal detectable amount of favipiravir per voxel was estimated to about 500 pmol. The results show that 19F photo-CIDNP is a promising method for the non-invasive detection of suitable 19F-containing drugs and other compounds with very low levels of the substance.


Signal enhancement as a function of illumination time
The 1 H FID representing almost entirely the solvent signal is recorded over the whole sample leading to a non-expontial decay due the increased B0-field inhomogeneity.The 19 F FID stems from a much smaller field-of-view exhibiting a more homogeneous B0-field.T2* for 1 H was estimated by the vendor-provided shim-sequence to about 32 ms which is within the same range as T2* of the 19 F signal.The lower panels show the according resonance lines.

Spectroscopy with and without illumination
Fig. SI 3 shows spectroscopic measurements with and without illumination.Because of the low concentration of favipiravir non-illuminated measurements required hours to days.Each data acquisition of a non-illuminated spectrum was preceded by the acquisition of at least one hyperpolarized spectrum (32 averages) to check the integrity of the system (illumination duration 4 sec).For non-illuminated spectra, measurement times of up to 32 h were required.Bandwidth was 1 kHz or 4 kHz, sampling rate 16k, TR 10 s (including 4 s illumination time for measurements of hyperpolarized spectra).Interestingly, the spectra of the non-illuminated samples showed a second signal, which was detected already after day one.This signal was shifted to lower frequencies by approximately 44 ppm (1000 Hz) and was not seen in spectra under illumination.The amplitude of this second signal (signal 2 in Figure SI 3c-d In contrast, the hyperpolarized signal changed little until day 7.Only from day 9 on, the sample showed a strong decrease of the hyperpolarized signal (Figure 3e).
Thus, unlike 3-fluorotryrosine [1], the hyperpolarizable form of favipiravir appears to be less stable when stored in the magnet at 303 K, even with little to moderate illumination (hyperpolarized spectra were recorded between the non-illuminated spectra).The only slightly varying degree of the hyperpolarization between day 0 and day 7 suggests a relatively constant amount of hyperpolarizable favipiravir.We have not yet determined the molecular structure of the second molecule because of the necessary difficult and lengthy experiments, but a slow transition of the original hyperpolarizable enol form of favipiravir [2] to the ketone form may provide an explanation [3] (see main text discussion section).
The observation that the signal 2 increased while the hyperpolarized signal remained relatively constant may be explained by the lower riboflavin concentration compared with the favipiravir concentration: sufficiently hyperpolarizable favipiravir molecules remained even when the concentration of the enol form was greatly reduced (day 4-7).Further experiments, such as changing the pH, are needed to gain additional insight into possible mechanisms of the transition between enol and ketone forms.For the ease of the reader, figure 7 of the main text is shown here as figure SI 4.

Figure SI 4:
The 19 F of the enol form is hyperpolarizable due to the heteroaromatic ring system and the OH group.In the ketone form, the ring system has undergone a transition and the H nucleus is now bound to the N prohibiting the transfer of the hyperpolarization to the 19 F. Therefore, no 19 F hyperpolarization could be detected in the ketone form.

Figure SI 1 .Figure SI 2 .
Figure SI 1. SNR (relative to the maximum) of photo-CIDNP-hyperpolarized 19 F signal of favipiravir as a function of illumination time.Four measurements were averaged for each illumination time.Delay between consecutive data acquisitions 11 s.Phase-corrected real values of the spectra were integrated and normalized to the values of the integrated 1 H spectrum.An illumination time of 4 s corresponded to approximately 80% of the maximum achievable signal enhancement.2.T2* times of 1 H and 19 F

Figure SI 3 .
Figure SI 3. Long-term time measurements of the 19 F signal in favipiravir.a-d: Absolute values of selected spectra (complete bandwidth) show the hyperpolarized 19 F signal (1 in Figure SI 3a) in the illuminated sample.The corresponding 19 F signal (1) in the non-illuminated sample is slowly decreasing while an additional increasing second signal (2) is seen.Signal 2 is not present in the spectrum of the illuminated sample (a).a: Illuminated sample at day 0 (32 averages).b: Non-illuminated sample at day 0 (2048 averages).c: Non-illuminated sample at day 4 (4096 averages).d: Non-illuminated sample at day 9 (3553 averages). 19F signals were re-calibrated to the first measurement under illumination with respect to bandwidth and signal fluctuations using the 1 H signal, which were due to remaining minimal temperature fluctuations in the temperature-stabilized magnet.e, f: Integrated phase-corrected real values of the 19 F spectra for illuminated samples (e) and non-illuminated samples (f) as a function of time.Black dots correspond to signal 1and red dots to signal 2.