Introduction

Continuous manufacturing

The conventional pharmaceutical manufacturing process is a batch process in which individual unit operations (i.e., feeding, powder blending/mixing, granulating, drying, coating) are performed. The batch process is flexible because the manufacturing process can be easily reconfigured. It can be worked empirically according to basic principles, and it is simple to track when a pharmaceutical quality problem occurs (Ierapetritou et al. 2016). However, because quality control through off-line testing is performed among these individual unit operations, in a batch process, the manufacturing line is often stopped and restarted, and the manufacturing line is also cleaned during this time (Helal et al. 2019). In addition, it is essential that equipment and space are sufficient to handle and store intermediates emitted from each unit process. Therefore, the batch process requires a large amount of human resources and manufacturing costs, which have limitations such as batch-to-batch variability due to human factor intervention and technical difficulties in scaling the batch to pilot scale (Ierapetritou et al. 2016).

In contrast, continuous manufacturing follows the “one-in, one-out” principle, whereby materials are continuously added, and the final product is discharged at a constant flow rate, with no starting and stopping between each unit operation. Continuous manufacturing can therefore be thought of as a new pharmaceutical manufacturing process with a new concept that can increase batch size over time. Therefore, continuous manufacturing does not require constant updating of equipment required for scale-up research. It is possible to operate the process with only a small number of workers, or it can be unmanned during 24-h processes. In this process, real-time process analytical technology (PAT) is used to ensure the quality of the finished product (Fig. 1). Process analytical technology is a system that designs, analyzes, and controls pharmaceutical manufacturing by timely measuring of raw materials and in-process materials and gauging critical process parameters and performance to ensure final drug product quality (ICH 2009). This manufacturing method can improve manufacturing efficiency, save both time and money, and can easily be scaled up as the understanding of the process increases (Lee et al. 2015; Matsunami et al. 2018).

Fig. 1
figure 1

Simple diagram showing batch unit manufacturing and continuous process manufacturing for solid oral dosage forms (Lee et al. 2015; Vanhoorne et al. 2020)

Full size image

The size of a batch produced by continuous manufacturing can be defined as the quantity of output material, the quantity of input material, and run time at a defined mass flow rate (ICH 2021). Also, in general terms, the size of a production batch for continuous manufacturing equipment is defined as the unit time (Allison et al. 2015), and because it has fewer scale-up issues, the continuous production equipment can be used at the same scale for drug development, pilot research, clinical trials, and commercial production. Consequently, the overall development time of products developed by continuous manufacturing is significantly reduced, and their time to market is significantly accelerated. In particular, the process of developing a new drug to market occurs over a long period and is hugely expensive. If a pharmaceutical manufacturing process is introduced through a continuous process when developing a new drug, a longer exclusivity period can be enjoyed even if the patent expires. Because the continuous production process is not a technology that anyone dares to imitate, even if the patent expires, you can enjoy a longer patent exclusivity period. So, it means that companies that apply these technologies can reap enormous economic benefits (Fonteyn et al. 2015; Shaver et al. 2011). In addition, there is a possibility of using continuous processes in the extension of patent rights, and gaining a competitive advantage through cost reduction and quality improvement, if continuous manufacturing is used in the development of generic drugs (Ierapetritou et al. 2016). Other advantages of the integrated continuous manufacturing platform are shown in Table 1.

Table 1 Characteristics and advantages of continuous manufacturing (Fonteyne et al. 2015; Lee et al. 2015; Burcham et al. 2018; Vanhoorne et al. 2020; Wahlich 2021)
Full size table

Quality assurance in continuous manufacturing

Previously, there was no specific mention of a continuous process in the ICH guidelines; it was discussed only as an improved manufacturing management method for the actual implementation of quality by design (QbD). However, a draft of specific content for continuous manufacturing was recently mentioned in ICH Q13, released in 2021, which can be used as a reference for detailed guidance on continuous manufacturing (Table 2). This means that in order for us to develop a successful continuous manufacturing technology, we need to understand ICH Q13 and establish management strategies based on the definitions and principles presented in ICH Q8–Q12.

Table 2 ICH guidelines, including continuous manufacturing contents (ICH 2005; ICH 2008, 2009, 2019, 2021, 2023; Wahlich 2021)
Full size table

QbD is a systematic approach to drug development, with the overarching goal of providing a consistent supply of drugs with guaranteed critical quality attributes (CQA). The continuous process was understood as a pharmaceutical manufacturing method consistent with the QbD principle, as it can improve drug quality by understanding the production process and ensuring CQA (Nasr et al. 2017). In this regard, examples of the current status of commercialized continuous processes can also be confirmed (Testa et al. 2020).

As the continuous process application develops, PAT is utilized to monitor the CQA of intermediates and processes in real time, making it possible to construct a system that designs, analyzes, and controls the manufacturing process. Therefore, each CQA will need to be included in the appropriate range and limit to ensure the quality of the drug product. The implementation of PAT requires utilizing the following methods: identification of traceability and residence time distribution (RTD) through the PAT framework; continuous monitoring of material properties, critical process parameters (CPP), and real-time release testing (RTRT); and maintaining the process and CQA within the operating range by controlling the drug manufacturing process operation in real-time (Zhong et al. 2020). RTD is defined as the time required for material transport and transformation; ICH Q13 explains that understanding this will be beneficial in material tracking, sampling, and the establishment of conversion strategies.

In addition, it is important to explore regulatory aspects when applying continuous processes in the pharmaceutical industry. ICH Q13, currently under discussion, provides an overview of current regulatory considerations. These considerations are overview of the process, quality control strategy for output, the definition of batch, process model, stability of manufactured drug product, process validation, pharmaceutical quality system (PQS), lifecycle management, and cross-referenced Common Technical Document data. If the manufacturing method is changed from a batch process to a continuous process, these data should be described. Table 3 is currently covered in the ICH Q13 draft. In addition, through major regulatory agencies such as FDA, MHRA, EMA, and PMDA, it is possible to understand the global regulatory environment for continuous processes and to stay up-to-date on support and regulatory status (Wahlich 2021).

Table 3 Regulatory considerations when applying continuous processes (ICH Q13, 2021)
Full size table

The pharmaceutical industry needs to develop a more efficient pharmaceutical manufacturing process, which could mean introducing the continuous process discussed in the ICH guidelines, along with the associated PAT implementation and management strategy development. Therefore, it is necessary to extensively review the PAT to determine what analytical equipment is used for a specific process, and what are the important process quality characteristics. Therefore, we will list and discuss the current state, scope of technology, and mode of application of PAT for commonly produced solid oral dosage forms, with a particular focus on the latest research in this area. We refer to previous reviews (Fonteyne et al. 2015; Vanhoorne et al. 2020; Zhong et al. 2020), and further supplement important information by referring to the continuous manufacturing process for the drug products listed in ICH Q13.

General manufacturing diagram of solid oral dosage forms

A generalized manufacturing process for a solid oral dosage form is shown in Fig. 2. The main unit processes include mixing, wet granulation, dry granulation, spray drying, compression, and coating. However, having a continuous flow of materials is also imperative. Therefore, a control strategy that is suitable for the continuous manufacturing of such a solid oral dosage form is necessary when developing a continuous process and management strategy, meaning that the CQA for each unit process needs to be identified. The various CQA for each unit process are shown in Table 4 and include content uniformity, which is an important CQA that is monitored at almost every step.

Fig. 2
figure 2

General manufacturing flow for solid oral pharmaceuticals

Full size image
Table 4 CQA for each unit process to be considered for continuous process monitoring
Full size table

Implementing continuous Process Analytical Technology in each unit process

The existing batch process was mainly conducted offline, while quality evaluation was performed on the intermediate and final products. Since this is a post-analysis quality evaluation, the manufacturing process cannot be immediately adjusted, so it is not suitable as an improved process quality control method. However, the PAT applications can measure the CQA related to various unit operations of a continuous process line in real-time (Zhong et al. 2020). Therefore, during the continuous process, CQA is evaluated in real-time via at-line, on-line, and in-line measurements according to the PAT industry guidelines as published by the US Food and Drug Administration (FDA). Each measurement method is described as follows (FDA 2004) (Fig. 3).

  • At-line: measurement where the sample is removed, isolated, and analyzed near the process stream.

  • On-line: measurement where the sample is diverted from the manufacturing process and could be returned to the process stream.

  • In-line: measurement where the sample remains in the process stream and can be either invasive or non-invasive.

Fig. 3
figure 3

Comparison of the At-line, On-line, and In-line PAT measurement methods

Full size image

The placement of the analyzer’s probe and sensor is an essential consideration when developing and establishing a continuous process control strategy. In this process, in-line measurement can be challenging because of potential contamination of the probe, difficulty in positioning the probe, and difficulty in defining the sampling volume of the powder (Sacher et al. 2022). Both probe and sensor should be positioned to measure the powder flow at a location that does not affect the flow in any way (Fonteyne et al. 2015). If the continuous process manufacturing equipment and PAT tools are appropriately designed or selected in consideration of these points, the process will be simplified, and process monitoring and material conversion can both be facilitated (ICH 2021).

Consequently, this review paper also identifies equipment and process parameters that affect process control based on ICH Q13’s continuous process management strategy. The impact on the material quality characteristics and the unit operating equipment is evaluated as CQA. A description of the PAT according to the manufacturing process flow of solid oral pharmaceuticals is detailed in the next section.

Continuous material streams

An essential element of any continuous process is to ensure a sufficiently constant mass flow (Engisch et al. 2012), since consistently and accurately supplying material from the unit process to the subsequent unit process is important. Therefore, a loss-in-weight (LIW) feeder system that analyzes injector mass flow and variability is typically used for all major routes during a continuous manufacturing process and is also utilized in all major routes of the continuous production lines of solid oral pharmaceuticals, such as direct compression, wet granulation, and dry granulation (Hanson 2018). This is particularly important because LIW feeders can generate fast dynamic disturbances, which are controlled through process monitoring (ICH 2021).

The ICH Q13 guideline states that it is essential to understand the flow of substances (process execution time and RTD) in individual units of equipment design in order to establish the integration process. In particular, because RTD represents the diffusion of material in a continuous process, it is important for the tracking of raw materials during process execution, and it allows downstream control or removal of the affected material in the entire continuous system (Engisch et al. 2016). A detailed overview of the principles and mechanisms of the continuous process feed step can be found in Blackshields et al. (2018).

Consequently, evaluating the homogeneity of the powder stream and the monitoring of the API concentration is conducted during the continuous material stream process. Monitoring techniques usable in continuous processes related to continuous material streams have been developed as follows. In addition, if the monitoring technology is closely intertwined with other major unit processes in the continuous manufacture of solid oral products, it is classified as one of the main unit processes, which are described later in more detail (Table 5).

Table 5 PAT tools and evaluable CQA used in continuous material streams
Full size table

The monitoring method for evaluating the flowability of powder is generally analyzed, in-line, in a LIW feeder. Additionally, a monitoring method using near-infrared spectroscopy (NIRS), which is the most commonly used technique for process analysis of continuous processes, is also widely used. In particular, as a result of comparing the NIRS monitoring system presented in (Destro et al. 2021) and high performance liquid chromatography (HPLC) and spectroscopic measurements by offline tests, successful in-line measurements can be made, such as active pharmaceutical ingredient (API) concentrations, which tend to have a high level of consistency. Therefore, as a recent research trend, it is in progress that a more detailed study on the influence of supply parameters and the identification and analysis of various CQA during the initial powder supply process. In addition, various combinations of these monitoring approaches are made. Therefore, it is expected that a hybrid form of latent variable modeling and state estimator for fault detection and control will exist from now on (Destro et al. 2020). In addition, an important element of PAT equipment for continuous material stream is the detection of all significant disturbances through high-frequency measurement methods. Therefore, a method of tracing raw materials using RTD data and identifying their characteristics is being developed as a form of corrective action when a disturbance starts, along with utilization of a diagnosis system (Engisch et al. 2016).

Continuous blending

Blend uniformity in the mixing process step is important since it is a prerequisite to ensuring the uniformity of the final drug content. Offline HPLC analysis is often used to evaluate general blend uniformity in the existing batch unit. However, this method has disadvantages such as possible sampling errors, process delays, and increased time and labor (Zhong et al. 2020). Therefore, the following PAT was used in continuous blending monitoring to overcome this limitation. Similarly in the continuous blending step, the RTD identified shows its impact on process dynamics and equipment design in the course of comparing APIs with tracers that have similar flow characteristics to APIs (ICH 2021). ICH Q13 shows that RTD characteristics can provide information on forward and backward blending and the degree of jamming, which can define material traceability and conversion strategies. Additionally, other scientifically validated theories can be used to establish process kinetics knowledge. Based on these characterizations, other CQAs besides blending uniformity are shown in Table 6.

Table 6 PAT tools and evaluable CQAs used in continuous blending
Full size table

A widely used representative process analysis method, as aforementioned mentioned, is NIR-PLS (partial least squares). In addition, RS (Raman spectroscopy) is a suitable alternative to monitor lower concentrations than NIRS. However, according to (Galata et al. 2021a, b), it is difficult to accurately quantify low concentrations (less than 2w/w%) of API in both of the aforementioned methods. Therefore, there is a need for a PAT tool that can accurately quantify the amount of low-concentration API in the continuous powder blending process.

Recent research is focusing on how PAT is being developed to evaluate the suitability of the current technology for drug development, including low-concentration API. For example, machine vision systems can quantify intensely colored APIs, even in low-density images, when viewed with a digital camera (Galata et al. 2021a, b). There is a study of chemical composition monitoring of metal–organic reaction products using an explosion-proof online NMR sensor with a sampling rate of 15 s. This was introduced as a method that allows access to unstable lithiated intermediates that cannot be calibrated using conventional HPLC analysis methods. In addition, NMR sensors are available as an on-line method for calibrating NIR spectrometers (Kern et al. 2019). Therefore, the transition to a continuous process requires ongoing research to develop PAT tools suitable for all types of API.

Wet granulation

Wet granulation is a process whereby granules are prepared using a liquid binder to improve the fluidity, homogeneity, and compressibility of the mixed powder before tablet compression (Suresh et al. 2017). Twin-screw wet granulators (Huang et al. 2010; Fonteyne et al. 2014a, b), fluid bed granulators (Chablani et al. 2011; El Hagrasy et al. 2013; Fonteyne et al. 2013), and high-shear mixers (Fonteyne et al. 2012; Kumar et al. 2013) are the main methods used during the continuous wet granulation process. The CQA of the wet granulation process includes the solid state, moisture content, and granule size distribution of API and excipients, which is particularly important since particle size, moisture content, and granule density can affect granule fluidity, compressibility, and stability (Fung et al. 2006; Närvänen et al. 2008; Ehlers et al. 2009; Gabbott et al. 2016). It should be noted that for wet granulation, controlling and evaluating the properties of the input material properties during the continuous process may be more important than the actual material specification properties typically considered in batch manufacturing. This is described in ICH Q13 in the section regarding the development of a continuous process control strategy (ICH 2021). Therefore, recognizing the importance to identify and manage the properties of the materials, PAT development should continue apace (Table 7).

Table 7 PAT tools and evaluable CQA used in wet granulation
Full size table

During the wet granulation process, the majority of PAT tools are used for the in-line monitoring of granule size distribution. In addition to NIRS and RS, various process analysis tools such as FBRM, SFV, imaging technology, and acoustic emission (AE) have been studied, and their applicability has been proven. However, although the NIRS tool is one of the most commonly used methods, it requires a relatively large amount of data for calibration and can sometimes produce inaccurate data because it can be affected by sensor fouling (Rantanen et al. 2015). In addition, the optical device used in the image processing system can become contaminated, leading to resolution problems when checking the particle size (Suresh et al. 2017). One of the more recently developed tools, AE, is presented as a non-destructive method for particle size adjustment. It is also suitable for the wet granulation process because it is relatively unaffected by pollution and dust and it has a low implementation cost (Abdulhussain et al. 2021). In addition, the particle size analyzer using a customized image analysis-camera also showed excellent similarity when compared with the results obtained offline without major drawbacks, and it was possible to identify the characteristics of the process material according to the CPP. Because the latest algorithm is applied in this process, this method can be applied in a role such as deep learning-based image segmentation in the future (Madarász et al. 2022). Therefore, today's PAT tool, such as AE, is also being touted as an improvement over the existing PAT tools, with the added advantage of being easy to implement during manufacturing while nullifying some of the disadvantages of the existing PAT tools.

Dry granulation

Dry granulation is a granulation process that can improve the flowability of powder by improving particle size or bulk density. However, unlike wet granulation, dry granulation is a relatively simple process with no need for liquid additives (Burcham et al. 2018). In the continuous dry granulation process, a roller compactor compresses the material and then mills it to form a ribbon. The CQA of a roller compactor includes ribbon density, moisture content, particle size distribution, and content uniformity, which is a factor that shows granular properties (Fonteyne et al. 2015).

NIRS, which is a quality control method most commonly used for the roller compaction process applied in the pharmaceutical industry, unlike in the past, tends to lose a large gap between the real-time monitoring value and the standard value at the current technology level (SAMANTA 2012). However, the applicability of new techniques has recently been re-verified due to the shortcomings of NIRS data preprocessing and the need for more sophisticated chemistry (Nasr et al. 2017). In a recent study, the temperature distribution of the ribbon was confirmed by installing a thermal imaging camera on the roller compactor, and the optimal relative temperature uniformity was ensured (Yu et al. 2022). Studies to identify the relationship between process parameters and material properties are being conducted along with the application of PAT. In addition, information on laser diffraction, microwave sensors, infrared thermography, spatial filtering technique (SFT), particle size analyzers, NIRS and microwave resonance sensing is shown in Table 8, whereas more details about the PATs applied in dry granulation, can be found in the paper by Dular Vovko et al. (2020).

Table 8 PAT tools and evaluable CQA used in dry granulation
Full size table

Spray drying

Spray drying is a process used to convert a fluid material into a dried particulate form. This process is often used before the direct compression process and has also been utilized in other product manufacturing fields such as for the drying of heat-sensitive materials (Coppi et al. 2002; Picot et al. 2004) and microencapsulation (I Ré 1998). During the spraying process, the CQAs are the particle size distribution, as well as the quantification of the solution composition and residual solvent used in the process. Related PAT tools are shown in Table 9, while a more detailed description of spray drying can be found in Ziaee et al. (2019).

Table 9 PAT tools and the evaluated CQAs used in spray drying
Full size table

The first unit operating step of spray drying is the preparation of the spray solution (Lee et al. 2019a, b). NIRS can measure the composition of the final spray-drying solution through real-time monitoring at this stage and can correct any problems with the spray volume is incorrect. In this way, the accuracy of the analysis has been proven. In addition, NIRS can be used during intermediate quality control for the quantification of residual solvents and the optimization of drying process cycles (Ikeda et al. 2022). Various studies have been comparatively validated, in-line, at-line, and offline, including the use of an in-line laser diffraction system to monitor particle size distribution, which affects the properties of both the intermediate and final products (Chan et al. 2008).

Compression

When using a tablet press machine during the tablet compression phase, the PAT is mainly applied to the powder feed frame of the tablet press machine. The CQA of this process is the content uniformity of the tablet, the content of the main ingredient, the tableting pressure, and the physical properties (Kim et al. 2021). The process is controlled through monitoring, which determines the quality of the tablets (Table 10).

Table 10 PAT tools and evaluable CQA used in compression
Full size table

In the case of NIRS in tablet pressing, the PLS model was designed to quantitatively analyze the concentration of API and its excipients, while a dissolution performance prediction model was also established (Pawar et al. 2016). RS can also be used to determine the CQA of intermediate and final tablets from the tablet press (Nagy et al. 2017), while Terahertz spectroscopy is able to detect tablet weight by measuring the effective refractive index (Bawuah et al. 2014). A report was also written on the combination of two NIRS probes on a tablet press feed frame (Pauli et al. 2019a, b, c): the first probe confirms the API content uniformity of the dried granules, while the second probe allows accurate monitoring of tablet content uniformity at various tableting rates up to 70,000 tablets/h. In addition, according to a recent study, a new NIR-SS hybrid method combining online NIR-PLS and titer soft sensor was successfully performed. This method utilizes the average data of the degree of movement according to the process stay time through a combination of two data streams, and it is possible to diagnose real-time online performance (Cogoni et al. 2021). This means that unsuitable tablets that do not conform to the CQA are distinguished through various combinations rather than through a single utilization of PAT. In addition, the feedback control loop is designed to respond immediately to problems in the process. The overall goal, therefore, should be to establish a system that optimizes current quantitative methods and facilitates improved control strategies for future PAT development in a continuous process.

Coating

The coating process allows for the sustained or controlled release of API, and the stability of tablets. It is an important process that determines the overall appearance of the final product (Carter et al. 2018). Conventional coating thickness measurement methods are gravimetric, dissolution-based, and conducted offline. They are also time-consuming and cannot always provide information relating to tablet uniformity, porosity, and cracking. The CQAs of the coating process include coating thickness, API content, and content uniformity. Examples of real-time PAT development to evaluate these factors are listed below in Table 11.

Table 11 PAT tools and evaluated CQAs used in coating
Full size table

Various new technologies are applied in the coating process analysis. These include NIRS, RS, SFV, and BARDS, which are spectrum technologies; TPI and OCT, which are imaging technologies; and sound technologies. Among these technologies, in-line OCT has been widely used in recent film coating research; it is a method that can monitor parameters such as tablet coating thickness and content uniformity in real-time using high-resolution imaging. Sacher et al. (2019) explain that OCT can detect defined values and deviations during the coating process faster and more accurately than conventional techniques. A recent study confirmed the high accuracy of tablet moisture content using a PLS model through manual wavelength selection (Shibayama et al. 2021).

The coating process has nevertheless proven to be arguably the most difficult unit operation to convert into a continuous process because of its complexity and its many parameters (Suzzi et al. 2010). In addition, tablets tend to relax after compression, so if the coating process proceeds too quickly, the coating may swell and affect the quality of the finished product. So the holding time before the tablet is coated is an essential process element. However, this holding time conflicts with the characteristics of the continuous process. Therefore, the continuous coating system is known to have developmental limitations meaning that there is a lot of potential for future research and development (Leane et al. 2018; Wahlich 2021). In addition, because the amount of data is not large in the coating stage, it will be necessary to introduce PAT through a selection of optimal hyper-parameters and confirmation of correlation in the future (Shibayama et al. 2021).

Conclusion

This review paper was written based on the management strategy for continuous manufacturing mentioned in the draft ICH Q13 guidelines and discussed the details of process monitoring and management among the elements of the control strategy. Real-time monitoring PAT enables us to establish a scientific understanding of the process and control strategies when implementing a continuous manufacturing process. In particular, NIRS and RS are traditionally widely used as tools for process analysis in the manufacturing process of solid oral dosage forms, and most CQA monitoring is possible using only these two tools. However, as mentioned in the text, both NIRS and RS have their disadvantages. An increasing number of studies are therefore evaluating whether existing tools can be improved upon while determining the applicability of real-time monitoring analysis techniques of new tools. In this process, some of the most recent articles covered in this review paper discusses the detailed analysis of various CQAs through multi-point measurement, and flexibly utilizes various PATs by mixing existing process analytical techniques. Therefore, using PAT in recent studies provides in-depth knowledge about PAT as a real-time monitoring technology.

In addition, the development processes of the majority of continuous manufacturing technologies often only concluded with a review of each unit process. It is therefore necessary to discuss more extensive in-line monitoring to apply and integrate technologies in more diverse unit processes. One of the most recent papers related to this content was written by (Sacher et al. 2021). In addition discussions about the continuous manufacturing process of solid oral pharmaceuticals as well as the formation of an end-to-end system from drug substance synthesis to final formulation are being reported (Wahlich 2021). Therefore, the exploration of real-time monitoring analysis technology utilized in the solid oral dosage form in this text can be referred to as a basic guideline for PAT trends. However, we also need to be aware of and follow the latest discussions about the broader continuous process.