This is a cautionary tale about the need for robust product characterization and release specifications for all cell therapy products.
Background
While our food often has a list of ingredients, our drugs don't. We rely on our regulatory agencies to rule on the safety of our drugs. These agencies require drug manufacturers to submit to them the composition of their therapeutic compounds and then to comply with the product specifications. It is this composition and these specifications which formed the basis of the clinical data evaluated by the agency and upon which the marketing approval is based. Any deviation from those specifications requires a submission to the regulatory agency for review. Any deviation without such a submission is punishable.
At the manufacturing site, as products come off the line they are subjected to a panel of product release tests to ensure each batch complies with the product specifications.
Specification compliance is a direct function of the consistency of the raw and ancillary materials, equipment, and operating procedures used in the manufacturing process.
Cell therapies present unique challenges when complying with this paradigm for several reasons only two of which I will mention here. Firstly, it is not possible to achieve the level of product purification as one might with other therapeutic products. Secondly, the product characterization is at a cellular rather than molecular level.
In cases where an autologous cell therapy is centrally manufactured, they are most often subjected to product release testing similar to that described above. One notable difference, particularly for fresh products, is that the products may be shipped to the clinic and even administered before the full panel of test results are obtained. This wold be considered highly unusual (if ever acceptable) with other types of products but is tolerated because of the time-sensitivity of these products and their high safety profile.
In the case of autologous cell therapy products produced at the bedside there is often not the same kind of product release discipline. Often the regulatory agencies deal with the product consistency and specification compliance issue by ensuring that the cell processing device used point-of-care is validated to ensure the cellular product output is always within a specified range shown to be clinically safe and effective.
The Varying Degree of Product Characterization/Specification of Autologous GTP Cell Therapy Products
However - and now I get to the point of this blog post - for cell-based products, procedures and/or devices/kits which are not mandated to be formally approved by a regulatory agency before they can be commercially marketed, there is no product specification rigor. Compliance with the Good Tissue Practice regulations and guidance is deemed to ensure safety. In the United States, cell-based products which are deemed to be "minimally manipulated" and intended for "homologous use" are typically allowed to go straight to market with no formal approval. Safety and clinical data is not required but is practically necessary to support physician adoption and, where applicable, reimbursement.
This means that for these products there is a great deal of variability in terms of how much rigor companies apply in characterizing their product and then ensuring that each batch complies with the specifications they themselves have determined to be safe and effective. Again, where such products are manufactured in a centralized facility the likelihood of some release testing is greater. However, those companies relying on a point-of-care processing kit or device business model that has not been deemed to require formal market approval, rarely (if ever) include product release testing.
The common criticism of these companies is that they simply do not know what they are injecting into patients because of the combination of the patient-to-patient donor variability, the lack of any disciplined product characterization or dosing studies, and the absence of any product release testing.
This criticism is not equally levied at all autologous GTP products or companies - even those relying on point-of-care processing. Of course some companies care and do a lot to try to ensure their product is well-characterized and that each batch complies with product specifications. This may involve the use of product release tests but can also involve the combination of pre-market research into the product characterization, safety, and dosing along with validation of the device/kit output. In this way a company can say that within a very small margin, the output will be within the product specifications the company knows is safe and efficacious.
However, in a rush to get their device/kit to market some companies appear to care very little about the cell product characterization, validation of the output of their device/kit, or tying this data to optimal dose.
More concerning are those companies that appear to provide such data but it is wrong or meaningless. What follows appears to potentially be a case study of precisely this problem.
The INCELL Study
This week I came across a fascinating white paper from Incell Corporation analyzing the output of adipose tissue processing kits of MediVet-America apparently demonstrating the inaccuracy of their cell counts (a common type of cell therapy product characterization) and calling into the question the cell count claims of Intellicell Biosciences (New York, NY) and Adistem (Hong Kong).
At the heart of the critique is the claim that the cell counting (product characterization) techniques employed by these companies counts as cells things (namely acellular micelles) which are not cells.
I encourage you to read the white paper in its entirety. They corresponding author told me to watch for one or more papers which they are preparing for submission to peer-reviewed publications shortly. Presumably these will rely on a larger data set and perhaps test other methodologies or technologies.
For the purposes of this blog, I've pulled what I believe are the most salient excerpts below:
Intrigued by the high cell numbers (5 to 20 million cells/gram) reported by kit/device manufacturers such as MediVet-America (Lexington, KY), Intellicell Biosciences (New York, NY), and Adistem, Ltd. (Hong Kong) in adipose stem cell therapy compared to other methods (e.g., Chung,Vidal, and Yoshimura), INCELL staff conducted a research study to investigate the high apparent yield of stem cells. This initial work was focused on SVF cells from the MediVet Kit, which is marketed to isolate adiposederived canine SVF and stem cells.
The cell yields reported for the Medivet Kits are five to more than ten times higher than the yields routinely obtained by INCELL from freshly harvested human or animal adipose tissue using our adipose tissue processing methods. These yields are also tenfold or higher than those reported in the literature by most academic researchers (Chung-canine, Vidal–equine, Yoshimura–human). Since these cell counts are used to support stem cell dosing recommendations and cell banking, it is important to better understand why the cell numbers are higher.
...
A comparative analytical study of three dog donors of adipose tissue was designed to evaluate the cell yields using the MediVet Kit as an example of this class of isolation system. All kit procedures were followed as per the instructions provided. A brief overview of the different cell counting methods used, and the resultant cell counts, observations and explanations of the results observed, are described below
....
This study shows that incorrect counting of adipose derived SVF cells and the subset of regenerative stem cells can subsequently result in inaccurate dosing, both in direct therapeutic applications and in cryostorage of cells for future use. The DAPI-hemocytometer cell count (manual) was considered the most accurate, but there are various sources of technical difficulties that can lead to incorrect cell numbers. The nature of adipose tissue itself with variability in dissociation by enzymatic digestion can all contribute to the outcomes. Fat tissue has a propensity to form acellular micelles and oils upon tissue disruption. Processing methods or reagents (e.g., Solution E or lecithins) can generate micelles that may be erroneously counted as cells. Autofluorescence and dye trapping or uptake by the micelles can lead to very high inaccurate cell counts when automated cell counting is used.
In this study the most inaccurate counting came from the Cellometer. When used according to kitrecommended guidelines and on-site training provided by Nexelcom for counting cells by the MediVet procedure, the Cellometer overstated the DAPI-hemocytometer cell count by up to 20X or more. The Coulter Counter protocols also led to incorrect, high cell numbers. Although the cell counts were still a bit high, the authors recommend the NucleoCounter, or similar equipment, as more acceptable for automated counting. The manual hemocytometer-DAPI method is the most accurate, but requires a highly experienced cell biologist or technician to make accurate counts and is not suitable for routine clinical use....Other companies also have claims of very high cell numbers when their processes are used. Adistem, like MediVet, states they add an emulsifying agent to their kits to assist in cell release, and they also use a light activation system. Their kits were not tested in this study but it is possible that the high cell numbers reported by Adistem are also incorrect and result from the same problems highlighted in this paper for the MediVet procedure. Ultrasonic energy, which is commonly used to manufacture micellular liposome structures and to disrupt and lyse cells, is another potentially problematic procedure for counting and verifying viable, regenerative cells. Intellicell 3uses ultrasonic energy to release cells from adipose tissue, and it is possible that resultant micelles or cell fragments contribute to the higher than expected cell numbers. This assumption could be verified with additional studies.
In summary, the authors caution that great care must be taken when using kits and automated cell counting for stem cell dosing and cryobanking of cells intended for clinical use. Overestimated cell numbers would be a major confounding source of variation when efficacy of stem cells injected are compared as doses based on cell number and when cryostored cells are aliquoted for use based onspecific cell numbers as a treatment dose. Hopefully, this study will lead to more reproducible counting and processing methods being reported in the literature, more inter-study comparability of cell doses to clinical outcomes, more industry diligence to support claims, and more accurate counting for dosing stem cell therapies to patients....
Chung D, Hayashi K, Toupadakis A, et al. Osteogenic proliferation and differentiation of canine bone marrow and adipose tissue derived mesenchymal stromal cells and the influence of hypoxia. Res Vet Sci, 2010; 92(1):66-75. Vidal MA, Kilroy GE, Lopez MJ, Johnson JR, Moore RM, Gimble JM. Characterization of equine adipose tissue-derived stromal cells: adipogenic and osteogenic capacity and comparison with bone marrow-derived mesenchymal stromal cells. Vet Surg, 2007; 36:613–622. Yoshimura K, Shigeura T, Matsumoto D, et al: Characterization of freshly isolated and cultured cells derived from the fatty and fluid portions of liposuction aspirate. J Cell Phys, 2006; 205:64-76.
In Conclusion
Despite some of their other challenges, Intellicell, MediVet-America, and AdiStem (and others) have scored credibility points with some of my colleagues who have been impressed by the fact that they have incorporated product release criterion and testing technologies into their business model where their peer companies have not bothered. This credibility may be quickly eroded if it turns out the results of their cell counts have been misleading. For now it is a word of caution to do your own due diligence and/or not to fall into a similar product development/characterization trap. Meanwhile, we will watch for the peer-reviewed papers.
5 comments:
Trying to be optimistic, minimally manipulated autologous cells used homologously are unlikely to do you any real harm, but they made do you no benefit either. Cell dose accuracy doesn’t’ normally appear to be critical, quite wide doses often show little measureable difference in clinical effect; but dose responses are seen. However, the differences observed are quite likely to alter the benefit, although it does say the claims are 10 fold higher than most published values, so the method is probably equivalent – just the user might dose differently as a result – but then has anyone done dose-finding studies anyhow? There is a general attitude in cell transplantation that more is better, give as much as you can: so here that would probably result in the same dose being given in reality (although assumed to be 10x). For a medicinal cell therapy this would be completely unacceptable, but then the cell content method would need to be thoroughly validated – it is hard to believe any validation was undertaken for the cellorometer and SVF.
Very interesting, and a little worrying, especially that device regulation may not be adequately evaluating analytical functions.
Unfortunately Christopher Bravery (regulatory consultant extraordinaire) was unable to post the entirety of his comment so he emailed it to me. With his permission, it appears below in two parts (a very worthy and worthwhile response).
___
Interesting story, and some lessons to learn in there regarding appropriate methodology and the need for validation (seems unlikely this was undertaken correctly, if at all).
However, some comments/perspectives and minor corrections.
Firstly, actually the ingredients of medicines are stated (in the EU at least, but equivalent information is available in a slightly different form in the US) in the summary of product characteristics (SPC or SmPC in the EU) lists all excipients (not amounts) and the amount of the active substance/s is included. So not all the ingredients (raw materials) that are used to manufacture the medicine, but those present. There are of course residual raw materials (impurities), but these have to be controlled, as low as possible and shown to be demonstrated to be acceptably safe.
Sometimes there is mention of materials used in the manufacture, where these might be important to know clinically, for instance the Apligraf label states “Product manufacture also includes reagents derived from animal materials including bovine pituitary extract.” Indeed I might argue the information provided is more reliable than for food.
Although in broad terms it is correct that most medicine active substances are very pure (although the final product may not be due to necessary excipients), it is not perhaps as uncommon as might be implied here that active substances are heterogeneous and poorly defined; good examples are medicinal blood products e.g. normal human Ig, polyclonal antisera etc. I’ve come across a fair number of other complex and difficult to define biologics – they are as big a headache as cell products.
What your discussion does pull out is the difference between a medicine and a cell or issue transplant (GTP’s only), because a transplant is minimally manipulated and used for the same essential function, it is an extension of surgery and left to the judgement of the surgeon. The public health concern (GTP) here is not practice of medicine (surgery) but protection of the donor and control of disease transmission – so donor consent, appropriate donor (adventitious agent) testing, and suitable facilities for any minimal manipulations. So legally there is no need to perform extensive testing on the cell transplant; and in general terms where is the risk?
Of course this also means no assessment of efficacy/effectiveness is mandated, this is left to practice of medicine to decide whether they undertake trails or not – or the device manufacturer if they are selling a device (may be required for device registration too of course). Which I guess brings us to the device, here the approach to regulation doesn’t appear to require detailed characterisation of the result, but you would have thought it would evaluate the ability of a cell counting device to count cells accurately.
I’m speculating now, and I confess to having some prejudices here, but I think this could have resulted from the tendency to regulate devices through standards, meet a set of standards and the device is fit for purpose. Standards work where things are predictable, and only if the scope of their use is carefully defined and adhered to.
So if they followed a standard for the cell counting system that didn’t anticipate micelles, they could well have complied with it. Which loops us back to the lack of need for characterisation for the biological material being processed. So perhaps a flaw in the system, although in this situation the responsibility lies with the surgeon and/or the device manufacturer where the regulation does not cover this.
(continued below)
(continued from above)
It could be said this blog demonstrates that where a point of care device is actually making a medicinal product, e.g. non-homologous use of minimally manipulated cells, regulation of devices is probably not the right regulatory mechanism, or would need changes. Many of these devices are evolving further to the point they can more than minimally manipulate the cells, which poses all sorts of problems since the legislative framework didn’t anticipate this.
Trying to be optimistic, minimally manipulated autologous cells used homologously are unlikely to do you any real harm, but they made do you no benefit either. Cell dose accuracy doesn’t’ normally appear to be critical, quite wide doses often show little measureable difference in clinical effect; but dose responses are seen. However, the differences observed are quite likely to alter the benefit, although it does say the claims are 10 fold higher than most published values, so the method is probably equivalent – just the user might dose differently as a result – but then has anyone done dose-finding studies anyhow?
There is a general attitude in cell transplantation that more is better, give as much as you can: so here that would probably result in the same dose being given in reality (although assumed to be 10x). For a medicinal cell therapy this would be completely unacceptable, but then the cell content method would need to be thoroughly validated – it is hard to believe any validation was undertaken for the cellorometer and SVF.
Very interesting, and a little worrying, especially that device regulation may not be adequately evaluating analytical functions.
______
Christopher Bravery of Consulting on Advanced Biologics (London) www.advbiols.com
I didn't know taking medicines or drugs can be this dangerous. It is indeed very important that manufacturers should have proper guidance and rules in terms of dissemination of their products.
The problem is that the manufacturing & quality processes do not check the actual accuracy of the counting method. The quality protocol may start with " We take a sample of 10.000 cells, we seed it and make it grow for XX hours..". The process will take good care that all the steps are properly taken.. but it will not ensure that the initial 10.000 cells are exactly 10.000 (or 8.000, or 6.000)
I wrote an article about
Errors in Cell Counting which explains the problem of cell counting using manual method. Without the proper care, automated systems like the one described on the report, or flow cytometers can be much worse..
Oscar Bastidas – Celeromics Cell Count Devices
Post a Comment