All printlets were of a high resolution and white in colour, with no discolouration during or after printing (Fig. 1). The final pharma-ink consisted of only three components: 94% w/w Klucel JF, 5% w/w D-mannitol, and 1% w/w doxazosin mesylate. This simple pharma-ink reduces the need for additional excipients such as fillers, binders, lubricants, or disintegrants, which are often needed in conventional tabletting processes [50]. With fewer formulation components, there is also a reduced risk of any potential interactions between excipients themselves or with the body, which are not always inert as investigated by numerous studies [51,52,53,54,55]. All printlets were printed from one batch of powder feedstock (pharma-ink), meaning dose titration would be much simpler and quicker for a pharmacist that must print the medication in the form of pharmaceutical compounding. The user would only have to adjust the software settings for the dimensions of the printlet, without changing the pharma-ink which would require completely changing the DPE printhead or cleaning the system.

Various printlet sizes printed using the optimised pharma-ink: (a) From left to right: 6 × 3.6; 8 × 3.6; 10 × 3.6, (b) From left to right: 6 × 3.6; Ch 8 × 3.6; Ch 10 × 3.6

Based on the SEM images of the side of the printlet, the individual printed layers can be clearly seen, with no gaps or inconsistencies between the layers (Fig. 2). Consistent pharma-ink deposition results in a strong printlet, with a uniform density throughout. The high printlet mechanical strength was confirmed by the breaking force data, with the average breaking force of the smallest printlet (6 × 3.6 mm) being 483 ± 1 N, very close to the maximum value measurable by the hardness tester (500 N). To ensure a similar extended-release profile for all printlets, the absence of gaps during printing is also important, leading to a predictable drug release profile.

SEM images of (a) side view of all deposited layers in the printlet and (b) zoomed in view of each deposited layer

DPE 3DP successfully printed all cylindrical printlets with minimal variation in the height, width, and channel diameter where applicable (Table 2). All printlet dimensions were close to the theoretical dimensions previously shown in Table 1. The printlet masses showed more variation and larger SD values, which can be explained by the variability in the amount of material being pushed out of the printhead during printing. To alleviate the variation in printlet mass, a force feeder could be added to the DPE hopper, which would add an additional mixing feature to the system as the powder is heated to ensure a consistent flow through the screw. This has already been done by Rosch et al., whereby a wire was shaped into an open noose and added to the coupling element, constantly rotating to ensure the powder is moved during printing [56]. Since all printlets were printed from the same batch of pharma-ink powder feed, the drug loading was calculated by dissolving printlets at random from different batches. Based on three different printlets, the drug loading was found to be 0.97 ± 0.01% w/w, close to the theoretical loading of 1% w/w.

To ensure no degradation has occurred during printing, TGA analysis was also conducted on all components of the pharma-ink (Fig. 3). D-mannitol exhibited the lowest degradation temperature onset of 239 °C, followed by doxazosin at 277 °C, and Klucel JF at 285 °C. Both Klucel JF and doxazosin have a ∼ 3% weight loss at around 80 °C, which is attributed to water evaporation, most likely due to some moisture presence in the samples. Printing of all printlets was carried out at 170 °C, which is well below the degradation temperatures of all components, confirming that DPE 3DP led to no drug or excipient degradation.

TGA thermogram of all pharma-ink components to assess degradation

Differential scanning calorimetry (DSC) was first carried out to assess the solid-state structure of the API on its own, as it is known in the literature to contain many polymorphs (Fig. 4) [57,58,59]. When zoomed into the thermogram, many thermal events are taking place in the API sample. Consecutive melting (endothermic event) and crystallisation (exothermic event) occur at different temperatures, indicating that the sample has a mixture of five polymorphs. The first polymorph (and the least stable one) melts at a Tmax of 230 °C, and recrystallises at a Tmax of 236 °C. This is then followed by successive melting and recrystallisation of each polymorph within the sample, until the last polymorph melts and recrystallises into the most stable polymorph at Tmax 280 °C. The first polymorph melting occurs at 230 °C, well above the printing temperature used (170 °C), meaning recrystallisation or melting should not occur during printing.

Zoomed in (200–300 °C) DSC thermogram of doxazosin mesylate and its mixture of polymorphs. All indicated temperatures are presented as Tmax

Differential scanning calorimetry (DSC) was then carried out to assess the solid-state structure of the API in the pharma-ink before printing and in the printlets (Fig. 5). As the drug loading in the pharma-ink (1% w/w) is below the sensitivity of most analytical techniques, information on the solid-state structure of doxazosin mesylate was going to be difficult to assess. Therefore, another pharma-ink with a higher drug loading of 10% w/w was made, by reducing the Klucel JF component quantity but keeping the D-mannitol (4% w/w) the same. This would ensure that the drug loading is above the sensitivity of all analytical techniques, and it can be assumed that whatever changes occur to the higher drug loading printlet, also happen to the 1% w/w printlet.

When comparing the DSC thermograms of the 1% w/w pharma-ink and printlet, no distinct melting endotherms for the drug can be seen in both, as expected due to the sensitivity limit of the instrument. D-mannitol shows a sharp endothermic peak at Tmax 167 °C, corresponding to its melting point which has also been confirmed by the literature [60]. This endotherm is present in all pharma-inks and printlets but is smaller, most likely due to the small quantity (4% w/w). The presence of D-mannitol endotherm in both printlets (1 and 10% w/w) also confirms that no degradation has occurred during printing. Despite increasing the doxazosin content to 10% w/w, no melting endotherm or crystallisation exotherm can be seen in the 10% pharma-ink or printlet. The first melting endotherm in the doxazosin sample at 230 °C is very small and may be masked by the large quantity of polymer in both the 10% w/w pharma-ink and printlet. In addition, the first polymorph melts at 230 °C but cannot recrystallise into a more stable form as it is now dispersed within the polymeric matrix, which is likely due to most of the formulation being made up of the polymer (86% w/w). Both the pharma-ink and printlet al.so exhibit noise at 275 °C, which is most likely degradation, confirmed by the TGA thermogram of doxazosin.

DSC thermogram of all pharma-ink components, as well as the 1% w/w and 10% w/w pharma-inks and printlets

Similarly to DSC, it was also difficult to assess the solid-state structure of doxazosin using the diffraction patterns of XRPD. As seen in Fig. 6, doxazosin does not have sharp intense peaks in the diffractogram. This means it cannot be distinguished from noise or other excipients in the formulation such as mannitol, which is made up of sharp and intense crystalline peaks. The diffraction peaks of doxazosin and mannitol also overlap at 14°, 21°, 23°, and at 25.5°, making it harder to isolate and detect any crystalline doxazosin specific peaks. This is evident in the 10% pharma-inks and printlets, where a small peak present in both the pharma-ink and printlet at 23° could correspond to doxazosin or mannitol. The most intense peak of doxazosin at 5° is also not seen in either the pharma-ink or printlet, giving no clear insight into the solid-state structure of doxazosin.

XRPD diffractograms of all pharma-ink components in the formulation, as well as the 1% and 10% w/w pharma-inks and printlets

On the other hand, FTIR analysis seemed to provide more information on the solid-state structure of doxazosin in the pharma-ink and printlet (Fig. 7). Crystalline doxazosin exhibits two broad peaks at 3357 and 3180 cm− 1, corresponding to the N-H stretching of the aromatic amine. Other characteristic drug peaks include the N-H amine bending at 1595 cm− 1, C-O stretching at 1168 cm− 1, and a sharp C-O peak at 1044 cm− 1 [61, 62]. Significant peaks do not overlap with the other components (Klucel JF and mannitol), making it easier to determine the solid-state structure of the drug. Again, 10% w/w pharma-ink and printlets were analysed and clear differences can be seen when compared to the 1% w/w formulations. Both the 1% w/w pharma-ink and printlet FTIR spectra look identical, again a result of the sensitivity of the technique. On the other hand, a clear difference can be seen between the 10% w/w pharma-ink and printlet. The two small N-H stretching peaks are present in the pharma-ink but disappear in the printlet, suggesting the drug is amorphous in the printlet. This is the case for all the characteristic peaks of doxazosin, which are present in the 10% w/w pharma-ink but are not visible in the 10% w/w printlet. This can be explained by the fact that amorphous materials often exhibit broader and less intense peaks, due to the lack of long-range order [63]. Therefore, some peaks present in the crystalline state may disappear in the amorphous form due to the disrupted molecular symmetry. Based on all the analytical date of the pharma-inks and printlets, it can be stated that the metastable polymorph of doxazosin is crystalline in the pharma-ink but becomes molecularly dispersed within the polymeric matrix during extrusion printing and forms an amorphous solid dispersion in the 1% w/w printlets.

FTIR spectra of all pharma-ink components, as well as the 1% and 10% pharma-inks and printlets. Lines correspond to significant peaks in the crystalline doxazosin sample

After printing, in vitro dissolution testing was conducted to assess the release profiles of each printlet. The three non-channel printlets, which have varying SA: V ratios (6 × 3.6–1.22 mm− 1; 8 × 3.6–1.06 mm− 1; 10 × 3.6–0.96 mm− 1) are expected to show varying release profiles. This is clear in the dissolution data (Fig. 8), where distinct differences are observed between the larger printlets (8 × 3.6 and 10 × 3.6) and the smallest printlet (6 × 3.6). Numerous studies have already evaluated the effect of SA: V ratio on the release profile of the drug in 3D printed dosage forms, showing that it is a critical parameter in the prediction of drug release [64,65,66,67,68,69,70,71]. However, all studies have been carried out on SSE or FDM 3DP, and currently no work has been done to assess its effect on DPE 3D printed formulations.

In vitro doxazosin release profiles of all printlets printed from the same batch with varying SA: V ratios (n = 3)

HPC is a swellable polymer, exhibiting typical hydrogel behaviour when in contact with a solution, with higher molecular weight HPCs showing greater swellability [72]. Printlets with a larger initial SA: V ratio (6 × 3.6) likely undergo more rapid and uniform swelling, exposing more surface area to the dissolution medium earlier on compared to larger printlets (10 × 3.6). In the first 60 min of dissolution, the smallest 6 × 3.6 printlet exhibits a burst release of doxazosin, as any drug present in the outer layer dissolves quickly. As the HPC swells and forms a gel layer over time, drug release slows and reaches a steady state, controlled by diffusion through the gel or erosion of the polymer. Tablets with smaller SA: V ratios (e.g. 10 × 3.6) may form thicker gel layers, which could slow the diffusion of doxazosin. In contrast, larger SA: V ratios (6 × 3.6) may result in thinner gel layers, leading to faster diffusion and dissolution. Despite differences in release kinetics, all formulations achieved ∼ 100% drug release within the experimental timeframe (16 h), suggesting that the SA: V ratio impacts the release kinetics but not the total drug dissolution.

To compare the release profiles more closely, f1 and f2 values were calculated for each printlet comparison. High f1 scores of 23.07 and 23.85 were obtained for the smallest printlet (6 × 3.6) when compared to the two larger non-channel printlets, 8 × 3.6 and 10 × 3.6, respectively (Table 3). An f1 score above 15 indicates that there are no similarities between the release profiles, which is consistent with the f2 scores. Both larger non-channel printlets had f2 values below 50 (44.77 and 43.52) when compared to the smallest printlet. An f2 value below 50 also corresponds to no similarities between the release profiles. In contrast, the f1 and f2 scores between the two larger non-channel printlets (8 × 3.6 vs. 10 × 3.6) were significantly better, with a low f1 score of 11.79 and a high f2 score of 63.86, meaning both release profiles are similar. This can be explained by the small difference of 0.1 mm− 1 in SA: V ratio, with the 8 × 3.6 being 1.06 mm− 1 and the 10 × 3.6 being 0.96 mm− 1. On the other hand, the SA: V ratio differences are larger for the larger printlets vs. 6 × 3.6 (Table 1), explaining the high f1 score and low f2 score.

To improve the similarity between the drug release profiles of all printlets from a single printing batch, single channels of varying sizes were introduced in the centre of the larger printlets. These channels were designed to achieve the same theoretical SA: V ratio (1.22 mm− 1) as the smallest printlet (Table 1). This modification led to improved alignment of the release profiles across all three printlets (Fig. 9), as evidenced by improved similarity indices in the f1 and f2 scores for all pairwise printlet comparisons (Table 3). The most significant improvement was observed between the larger 8 × 3.6 and 10 × 3.6 printlets, where the f2 score increased from 63.86 to 73.95, and the f1 score decreased from 11.79 to 6.54.

Drug release profiles of all printlets printed from the same batch with the same SA: V ratio through the introduction of channels (n = 3)

Although all printlets has the same theoretical SA: V ratio, the channel printlets exhibited slightly slower dissolution compared to the smallest 6 × 3.6 printlet. Notably, the channel printlets did not exhibit the burst release of doxazosin observed in the first 60 min for the smallest printlet. However, during the later stages of dissolution (240 min and beyond), all three printlets showed similar steady-state release profiles, outperforming the non-channel printlets manufactured before.

The absence of a burst release in the channel printlets can be attributed to the channels facilitating a more controlled hydration process. This likely allowed for a quicker and more uniform formation of the gel layer surrounding the printlet, thereby moderating drug release. Initially, the dissolution media may preferentially penetrate through the channel, slowing overall drug release across the matrix. In contrast, the smallest non-channel printlet (6 × 3.6) exhibited less controlled hydration, leading to delayed gel layer formation and therefore an initial burst release of the drug. Once the gel layer was formed for all three printlets, the drug release rate stabilised, and their in vitro release profiles converged, ultimately resulting in comparable overall release patterns.

While the f2 scores for both channel printlets improved (Table 3), they remained close to 50, which is the threshold for indicating similarity between release profiles. An f2 score near 50 indicates an average difference of ∼ 10% between the release profiles. This variation could be influenced by differences in density among the printlets or limitations in the overall printing resolution. Extrusion-based 3DP techniques typically exhibit lower resolution when compared to other technologies, such as resin-based printing or selective laser sintering. Resolution in extrusion-based systems is largely determined by the nozzle diameter [73, 74]. To further improve the f1 and f2 scores for the channel printlets (Fig. 9), a smaller nozzle diameter of 0.2 mm could be used to ensure a higher printing resolution and more accurate reproduction of the intended SA: V ratios.

Various drug release kinetic models were applied to the in vitro drug release profiles to determine the best-fitting release mechanism for the doxazosin printlets (Table S1). Among the non-channel printlets, the Hixson-Crowell model provided the best fit. This model assumes that dissolution occurs in planes parallel to the surface of the dosage form, with the surface area decreasing proportionally over time while the geometry remains constant [75, 76]. This correlates well with the uniform cylindrical structure of the non-channel printlets, which promotes consistent surface dissolution.

The Korsmeyer-Peppas model also demonstrated a strong fit for the non-channel printlets, with high R2 values close to 1. This model describes drug release from polymeric matrices, such as a hydrogel, where multiple mechanisms occur, such as diffusion, swelling, and erosion [77]. Notably, the differing SA: V ratios among the three non-channel printlets result in distinct release exponents (n) within the Korsmeyer-Peppas model (Table S2). For the cylindrical geometry of the smaller 6 × 3.6 and 8 × 3.6 printlets, the release mechanism corresponds to anomalous transport (0.45 < n < 0.89), indicating a combination of drug diffusion and polymer swelling [78]. In contrast, the largest printlet exhibits super case II transport (n > 0.89), where drug release is primarily driven by extensive polymer relaxation or swelling. The difference between these correspond to the differences in size and SA: V ratio, as a larger printlet has a greater volume and therefore a thicker polymer matrix, meaning there is a slower diffusion driven release.

The channel-containing printlets also showed a strong correlation with the Korsmeyer-Peppas model, achieving high R2 values of 0.9937 and 0.9834 (Table S1). Both Ch 8 × 3.6 and Ch 10 × 3.6 exhibited super case II transport (Table S2), attributed to the faster gel layer formation discussed previously, as well as enhanced hydration and greater polymer relaxation facilitated by the presence of channels. Differently to the non-channel printlets, both channel printlets do not fit well with the Hixson-Crowell model. The channels disrupt the overall symmetry of the printlets, leading to dynamic and irregular changes in total surface area during dissolution.

3DP of medicines can address sex-based differences in dosing and drug therapies by enabling highly customisable and personalised medications. As shown in this study, tailored dosages can be manufactured for each sex, accounting for the fact that males and females metabolise drugs differently due to variations in body weight, fat composition, enzyme activity, and hormone levels. With only modifications in the printing software, 3DP allows precise control over the amount of API in each dosage form. Sex differences are also shown to occur in the presence of some commonly used excipients in pharmaceutical formulations. 3DP can modify the formulation to minimise the use of non-inert excipients, reducing the risk of side effects or differences in drug absorption between individuals. In the future, personalised medicine from 3DP can integrate sex-based data from pharmacological research to refine formulations further. This approach ensures that medicine development addresses both male and female physiological needs effectively.