Personalized medicine uses various types of diagnostic information to improve outcomes and reduce healthcare costs by determining in advance whether a patient is likely to respond to a proposed treatment.

Of course, the best-known and earliest success of the use of diagnostics in personalized medicine relates to measuring HER2/neu amplification in breast cancer patients. Only those patients whose tumors carry this amplification respond to Herceptin (trastuzumab), and the drug's use is indicated only in such patients. Current study results indicate that treatment for a year reduces risk or recurrence by half in these patients.

 The cost of Herceptin treatment is about $3,000 per month and is easily justified because of the benefits it provides.

Getting patent protection for such diagnostics, however, recently became more difficult. The United States Patent and Trademark Office (USPTO) has made it difficult to obtain patent protection for certain types of personalized medicine technology. Unlike the test used for distinguishing Herceptin responders, i.e., a simple measurement of amplification of a single gene, other personalized medicine diagnostics require measuring gene or protein levels for a number of biomarkers. These measurements are used with sophisticated predictive modeling algorithms to generate a score associated with drug response or disease severity. One difficulty encountered in obtaining effective patent protection for this type of diagnostic arises from the ability to use various combinations of biomarkers to develop equivalent predictive models.

For example, while the expression of a handful of genes can be used in any particular predictive model, those genes can often be selected from a larger set of informative biomarkers. Consider a situation in which gene expression analysis identifies 30 genes whose expression significantly differs between identified groups of responsive and nonresponsive patients. Informative predictive models are generated using expression values for four or more genes. Consider further that many combinations of four genes out of the 30 provide good predictive power in the model. Ideally, a patent application would claim a method of determining the efficacy of the treatment by measuring the expression of four or more of the 30 genes and including those values in the predictive model. Such a patent claim encompasses 27,405 different combinations of genes. Because this is such a large number, a patent examiner may object that the large number of combinations creates an undue search burden for the examiner.

 Consequently, the examiner issues a "restriction requirement" that causes the applicant to choose one combination of, in this example, four or more genes whose expression values are used in the claimed method. Claims to other combinations would then need to be pursued in separate patent applications, raising the cost of protection to unaffordable levels.

This creates a difficult position for the patent applicant. If many combinations of four or more of the 30 genes are provided adequate predictive power, a patent for a single combination of four provides essentially useless patent protection. If restriction among each combination is required, the applicant is forced to choose which of the many combinations it developed and disclosed will it protect. The combinations that it cannot afford to protect can be used by competitors to develop essentially equivalent tests that are not protected by the applicant's patents. 

One potential solution to this problem may be revealed by additional statistical analysis of the identified genes. Such analysis may show that the 30 identified genes are not all equally predictive. Faced with a choice among many combinations, an applicant can select a subset of combinations that provide the most informative predictions for protection. Those combinations left unprotected, if carefully selected, would only allow competitors to produce inferior products unlikely to effectively compete in the marketplace with the patented tests.

Another potential solution relies on identifying subgroups of genes whose expression levels co-vary, making them natural substitutes for each other. Thus, the 30 genes may segregate into five subgroups—the members of which provide essentially equivalent information in the predictive model. Such analysis also may show that not all of the subgroups are equally predictive when input into the model. In such an instance, it may not be necessary to claim the product as testing any four of the 30 identified genes. Instead, the claim can be drafted in the form a combination of genes selected from these subgroups, using a common claim form known as a Markush group.

 A claim can be drafted as choosing five genes, one from each of the five subgroups. Additional analysis may show that if a gene from the most predictive subgroup, A, is chosen, expression of only two other subgroup genes is necessary for the model to be informative. Thus claims would also be drafted to choosing one of the genes in A and then two others from the various combinations available from the remaining 4 subgroups. Further analysis may show that even lacking a gene from the most predictive subgroup, using three or four genes from the remaining subgroups, B–E, yields an informative result. Therefore, claims should be drafted to combinations that use only genes from B–E. Those claims protect against competitors working around the ideal product using the most predictive genes.

The USPTO, however, has proposed new rules limiting the use of Markush groups. Under the new rules, 37 CFR 1.140 would be amended to require that all species encompassed by a claim that recites alternatives (i.e., a Markush group claim) "meet at least one of the following two conditions: (1) share a substantial feature essential for a common utility, or (2) are prima facie obvious over each other." Fed. Reg. 72(154):44997. The proposed method of diagnostics outlined above would meet even these new criteria. The grouping of claims would be done because they co-vary in response to a given criteria. As such they "share a substantial feature essential for a common utility." Also, because they co-vary, it could be argued that it is obvious to substitute one member of the group for the other.

Another option to address the undue search burden issue is for an applicant to request expedited examination from the USPTO. This route requires that the applicant provide a prior art search to the examiner. It allows for practitioners to develop search strategies for these new inventions and by submitting those search results to the USPTO, help the USPTO develop its search procedures for these inventions. The search strategy for such inventions can be divided into two broad categories—one where the genes being tested have previously been associated with the disease or therapy for which the predictive model was developed and another where the genes have not been previously associated with that disease or therapy. 

If the biomarkers have been previously associated with the disease or therapy, the most likely rejection from the USPTO would be that the invention is obvious. Obviousness rejections can be overcome by evidence of unexpected results. In the case of many predictive models, even if some or all of the biomarkers whose expression is being input into the model have been associated with that particular disease or therapy, the predictive model using multiple biomarkers is likely to be more powerful than predictions derived using any single biomarker. This gives a basis for arguing that the combination provides unexpected results, and therefore, is non-obvious.

In the case where the biomarkers have not previously been associated with the disease or therapy, the search strategy is simpler. The search should be structured to determine whether any of the claimed biomarkers have been associated with the disease or therapy for which the predictive model has been developed. A favorable search result should permit the issuance of broad claims to predictive models using combinations of novel biomarkers for predictive modeling of the therapy or disease with which they have previously not been associated.

Although there certainly are challenges associated with protecting these types of personalized medicine inventions, experienced patent counsel can enhance the ability of personalized medicine companies to capture appropriate claim scope.

 is a partner at Fenwick & West LLP, San Francisco, CA and 

 is an associate at Fenwick & West LLP, Mountain View,CA, 415.875.2413, 

1. Mayo Clinic Staff Herceptin: Novel therapy targets HER2-positive breast cancer. 2007 Oct 6.

2. Manual for patent examination and prosecution (MPEP) § 803.

Articles

Columns and Departments

There are several challenges associated with protecting patents for personalized medicines.

Protecting Patents for Personalized Medicine

This article is the third in a four-part series on essential statistical techniques for any scientist or engineer working in the biotechnology field. This installment deals with statistical methods for calculating confidence intervals, tolerance intervals, and capability analysis. The difference between the confidence interval and tolerance interval is explained. 

In the last installment, the concept of hypothesis testing was presented. In hypothesis testing, we must state the assumed value of the population parameter. In constructing intervals, on the other hand, we determine the range with a certain confidence for the true population parameter. This confidence level is usually set at 95%, though you might also see intervals that are 99% or 90%. A confidence interval is used to estimate the population mean or standard deviation. A tolerance interval is used to estimate the distribution of the individual values in the population. This distinction will become more apparent in the discussion of capability analysis. 

) the observed mean is from the hypothesized population mean can be determined by hypothesis testing. The conclusion is either the mean was statistically different or there was a lack of evidence to reject the null hypothesis. A similar analysis can be performed using confidence intervals. If the hypothesized mean fits within the confidence interval, it is an indication that the sample mean is not statistically different from the hypothesized mean. We will concentrate on the confidence interval when the population variance is unknown. When the population variance is unknown, the t-distribution, which takes into account the uncertainty in estimating the sample variance, is used. The t-distribution is tabled by confidence level and degrees of freedom. The degrees of freedom are the number of observations used to estimate the sample standard deviation minus one. The formula for the confidence interval is as follows: 

 is the t-value from the t-table with a confidence level of 1-α and 

 is the sample standard deviation used to calculate the sample mean, and 

 is the sample size used to estimate the mean and standard deviation. Using the protein concentration data from Part 2 of this article  series (

, June 2008), a confidence interval can be calculated. Table 1 shows the data and calculations for the confidence interval. If the theoretical concentration was thought to be 30, the 95% confidence interval shows that 30 is contained in the interval, therefore the mean of 31.70 is not different from the value of 30. 

Another way to look at the relationship between a confidence interval and hypothesis testing is to look at the hypothesis test to determine if the observed mean of 31.70 is statistically different from 28.47 (the lower confidence interval). The formula for the t-test would be: 

Table 1.  Data and calculations for the confidence interval 

The confidence interval for the two-sample condition is to develop an interval that determines if the difference between the means contains zero. The approach is similar to the one used in the above formula. Some textbooks use a pooled standard deviation (sp) when the population standard deviations are unknown, but assumed to be equal; and the samples sizes are small (under 30): 

A confidence interval covers a population parameter with a stated confidence, that is, a certain proportion of the time. There is also a way to cover a fixed proportion of the population with a stated confidence. Such an interval is called a 

. Statistical tolerance intervals are limits in which we expect a stated proportion of the individual values of the population to lie. For a sample mean (X mean) and sample standard deviation (

), the general formula for a tolerance interval is: 

 factor can be obtained from the appropriate table contained in ISO-16269-6 Statistical Interpretation of Data reference or calculated by statistical software.  

Revisiting the data from Table 1, the tolerance interval calculations for a 95% confidence and 95% coverage can be found in Table 2.  

Table  2.  Tolerance interval calculations for 95% confidence and 95% coverage 

The tolerance interval means we are 95% confident that 95% of the individual values in the population whose mean is 31.7 with a standard deviation of 3.07 will be in the range of 18.12 to 45.28.  

The assumption for confidence intervals and tolerance intervals is that the data are an independent random sample from a single population that is normally distributed. Typically, the true mean and standard deviation from the population are not known, and are therefore estimated from the sample. 

Process capability compares the output of a process to the specification limits. The comparison is made by forming the ratio of the spread between the process specifications (the specification "width") to the spread of the process values, as measured by process standard deviation. A capable process is one in which almost all the measurements fall inside the specification limits. Using confidence intervals or tolerance intervals, the percent of the values that would fall outside the specification can be calculated. This is best explained using an example.  

Assume we have a process whose specification is that the mean of 5 samples must be between 40 and 50. If we sampled from the process and observed a mean of 46 and standard deviation of 2 based on a sample of 5 units, the 95% confidence interval would be 42.9 to 49.1. The 99% confidence interval would be 41.0 to 51.0. Based on the 99% confidence interval, we can see that we would expect the mean to be outside the upper specification more than 0.5% of the time (because the entire probability of exceeding the interval is 1%, then the probability of exceeding just the upper interval is 0.5%).  

The exact value for the percent of the means of a sample size equal to 5 that would exceed the specification based on the observed mean and standard deviation can be calculated. This is accomplished by solving for the t-value in the confidence interval equation by setting the specification equal to the confidence interval limits. Because the mean is not centered in the specification, it will be necessary to break the problem into two parts, solving for the lower and upper limits individually, and then adding the probabilities together. Table 3 shows the results of these calculations. 

Table 3. Analysis of 99% confidence interval 

Confidence limits are limits in which we expect a given population parameter to fall. Statistical tolerance limits are limits in which we expect a stated proportion of the population to lie. Using the confidence interval and tolerance interval, specifications can be set that minimize the number of out-of-specification situations. A combination of tolerance intervals and confidence limits defines the overall process parameter (mean) and the distribution of the individuals.

 is the president of Statistical Outsourcing Services, Olney, MD, 301.325.3129, 

steven@statisticaloutsourcingservices.com

Statistical methods for calculating confidence intervals, tolerance intervals, and capability analysis to reduce out-of-specification situations.

Minimizing the Problem of OOS

A novel purification method was developed for recovering the pIDKE2 plasmid, which encodes a polyprotein encompassing amino acids 1–650 of the hepatitis C virus (HCV) polyprotein, from recombinant Escherichia coli. Bacterial cells were harvested and subjected to alkaline lysis. After centrifugation, the host contaminant RNA was removed from the clarified alkaline lysate using a highly loaded size-exclusion chromatography and the eluted fraction was applied to reverse-phase media: POROS R1 50. Finally, a second size-exclusion chromatography step was carried out to purify the plasmid DNA from other small molecular-weight contaminants. Analytical methods proved that the purified plasmid DNA had a purity of 95% after Sephacryl S1000. Plasmid identity was confirmed by restriction enzyme digestion. Biological activity of the purified plasmid was confirmed in vivo;  immunized mice developed a positive antibody response against all HCV structural antigens. This procedure offers an alternative to traditional methods that use organic reagents, mutagenic and toxic compounds, and animal-derived enzymes. Although the yields are lower when using this method, it is scalable and free of animal-derived substances and organic solvents.

Gene therapy and DNA immunization are becoming important alternatives for developing successful preventive and therapeutic treatments for many diseases. With thousands of people now receiving plasmid DNA (pDNA), it therefore must be produced by scalable manufacturing processes that meet stringent quality criteria in terms of purity, potency, efficacy, and safety. Plasmids are circular, double-stranded molecules that comprise approximately 1% of the total content of the host bacterial cell. They are normally isolated by an alkaline procedure, designed to disrupt the host cells and denature proteins and chromosomal DNA (chDNA), while preserving plasmid's structural integrity.

 Although commonly used in laboratories, pDNA  isolation methods that use elements such as organic reagents, mutagenic and toxic compounds, and animal-derived enzymes, are an additional concern for regulatory agencies, and therefore, must be avoided.

 Another challenge in purifying plasmids is eliminating contaminant cellular components from the host, generally 

, which induce immunological and biological responses. RNA removal presents a challenge in producing genetic therapeutics because of the similarity in chemical composition and structure with pDNA and its high abundance in crude plasmid preparation. 

This article describes a new pDNA purification method that uses a nonenzymatic approach to RNA removal based on size-exclusion chromatography on Sepharose CL 4B, using a buffer containing 1.5 M (NH

. The pDNA was pooled and directly applied on the POROS R1 50 media, which is a rapid alternative to conventional chromatography. For the last purification step, another size-exclusion chromatography method was chosen, because it can achieve two objectives in one operation—further purification and buffer exchange. Following these steps, the immunogenicity of the plasmid obtained was evaluated. Functionality of purified plasmids was confirmed 

 all immunized animals developed anti-HCV antibodies. 

The final process has proved to be generally applicable and can be used from early clinical phases to market supply. Although the yields are lower, this method is scalable and free of animal-derived substances and organic solvents.  

Chemicals were purchased from Merck (Whitehouse Station, NJ) and Sigma (St. Louis, MO). Ultrafiltration equipment and membranes were provided by Sartorius (Goettingen, Germany). Design Expert Version 6.06 (DX6) software was obtained from Stat-Ease, Inc. (Minneapolis, MN). 

Recombinant Co.120 and E1.340 were obtained from recombinant 

 with >85% purity, by a combination of washed-pellet procedures and gel-filtration chromatography.

 Co.120 comprises the first 120 aa of the HCV nucleocapsid protein. E1.340 encompasses aa 192–340 in the viral polyprotein. E2.680 comprises aa 384–680 of the HCV polyprotein and is obtained from recombinant 

, also by a combination of washed-pellet procedures and gel-filtration chromatography.

 DH10B cells harboring pIDKE2, a plasmid for DNA immunization expressing the first 650 aa of the HCV polyprotein from a genotype 1b-Cuban isolate, were grown over-night, at 37 °C, in 100-mL shake flasks containing 50 μg kanamycin/mL, at 250 rpm.

 was cultured using a complex media designed with a 50 g/L of yeast-extract concentration in fed-batch mode to produce pIDKE2 plasmid. 

Bacteria cell paste (typically 200 g) was suspended in 2.4 L of 61 mM glucose, 10 mM Tris-HCL, 50 mM EDTA pH 8. Lysis was performed by adding 2.4 L of 0.1 M NaOH and 1% w/v SDS, followed by incubation for 10 min at 4 °C. The lysate was neutralized with 2.4 L of 3 M potassium acetate for 10 min at 4 °C. Precipitated material including cell debris, most chDNA, and proteins were removed by centrifugation at 3,000 g for 10 min. The clarified cell lysate was concentrated five times on a tangential flow filtration (TFF) system (nominal molecular weight cutoff 100,000 kDa, 0.1 m

). All experiments were conducted maintaining constant transmembrane pressure (TMP) of 0.8 bar at 4 °C, using Sartocon Slice equipment from Sartorius. The lysate was loaded on Sepharose CL 4B in BPG 113 column to remove RNA using a 20 mM Tris-HCL, 3 mM EDTA, and 1.5 M (NH

 pH 7.2 buffer. The pDNA fraction was purified using  POROS R1 50 reverse-phase media and the elution was carried out in acetonitrile at 10% v/v. The DNA polishing step was carried out on Sephacryl S-1000 media in formulation buffer: 20 mM NaH

; 1 mM EDTA; and 30 mM NaCl pH 6.7 ± 0.2. Fractions containing more than 90% supercoiled plasmids were pooled and concentrated using PEG 8000 and NaCl, following methods published by Lis and Schleif.

Flow-through and eluted fractions were precipitated by 1 volume of 2-propanol for 15 min on ice and centrifuged at 14,000 x g for 10 min at 4 °C. The pellet was washed with 70% ethanol and centrifuged again at 4,000 x g for 10 min. The pellet was air-dried for 5–10 min and dissolved in 50 μl of 10 mM Tris-HCl pH 8.5. Fractions prepared as previously described were tested for restriction digestion analysis by transformation experiments and analyzed by 0.8% agarose gel electrophoresis. Yield and pDNA purity were determined by measuring absorbance at 260 (A260) and 280 (A280) and expressed as A260/A280 ratio in a low-salt buffer. As a positive control, supercoiled pDNA purified by the Horn method was used.

BALB/c (H-2d) female mice, 6 to 8 weeks old and weighing 18–20 g were purchased from CENPALAB (Havana, Cuba) and used for all 

 studies. Groups of 22 animals were immunized with the mixture of Co.120 (10 μg) plus plasmid pIDKE2 (100 μg), obtained using the method described by Horn

 or the method described here. Mice were injected intramuscularly at 0, 3, 7, 12, and 16 weeks in the quadriceps muscle with 100-μL final volume of 0.9% NaCl-based solution. Normal saline was administered to negative control group under similar conditions. Blood samples were collected 17 weeks after primary immunization from retro-orbital sinus and the sera were analyzed for antibodies (Ab) to HCV Co.120, E1.340, and E2.680 proteins. 

To detect murine anti-HCV antibodies, 96-well microtiter plates (Costar, Cambridge, MA) were coated with 100 μL of recombinant proteins Co.120, E1.340, or E2.680 (10 μg/mL) diluted in coating buffer (50 mM carbonate buffer, pH 9.6) and incubated for 12 h at 4 °C

 The wells were washed three times with 0.05% v/v Tween 20 in phosphate buffered saline pH 7.5 (PBST) and blocked with 200 μl of PBST containing 1% w/v skim milk (Oxoid Ltd, England) for 1 h at 28 °C. After three washes with PBST, 100 μL of serial double dilutions of individual mouse sera in PBST were added and incubated at 37 °C for 1 h. The plates were washed three times with PBST, and 100 μL of horseradish peroxidase-conjugated goat anti-mouse IgG (Amersham, UK) diluted 1: 25,000 was added at 37 °C for 1 h. Positive reactions were visualized with o-phenylenediamine (Sigma) in 0.1 M citric acid, 0.2 M NaH

 pH 5.0, and 0.015% v/v hydrogen peroxide as substrate, and the reaction was stopped with 50 μL of 2.5 M H

. Optical density (OD) was determined at 492 nm using a plate reader (SensIdent Scan, Merck, Ger-many).

The cut-off value to consider a positive antibody response was established as twice the mean OD 492 nm of the negative control sera. 

In the alkaline environment, the chDNA is denatured, while within a pH range at 12–12.3 pIDKE2 plasmid remains double stranded. During neutralization, the pH of the solution is reduced to a value close to 5.5 by adding potassium acetate. The change in physicochemical conditions of the solution causes the renaturation and flocculation of the chDNA, as well as the precipitation of protein-SDS complexes and cell-wall debris. The insoluble material can be separated from the liquid containing the pIDKE2 plasmid by centrifugation. Once the pellet is separated, the clarified alkaline lysate containing the pIDKE2 plasmid is concentrated 5-fold using a TFF system. This step, however, is not enough to remove all the RNA (Figure 1, Lane 4). On the other hand, the 260 nm/280 nm absorbance ratio was 1.8, which is in the range of the purity parameter regulated for a drug substance containing a purified pDNA.

 RNA removal can be performed  enzymatically using RNase or by selective precipitation using Ca2+; NH4+, or Mg2+ ions or polyethylene glycol.

 Depending on the qualitative and quantitative composition of the sample, this polymer involves the risk of co-precipitation of DNA and needs to be removed before subsequent ion-exchange steps. RNA removal also can be achieved by size-exclusion chromatography using Sepharose 6 Fast Flow, which recovers 92% of the pDNA. In this process, RNA was removed from the concentrated cell lysate during the first chromatographic step on Sepharose CL-4B recovering 91% of the pDNA by group separation in the presence of buffer containing 1.5 M (NH

 The use of high salt concentration is beneficial because of the different hydrodynamic size of doubled-stranded pDNA compared with small single-strand content such as RNA and other nucleic acids impurities. Therefore, pDNA eluted in the void volume can be clearly separated from RNA (Figure 1, Lane 5). The elution profile, obtained during the pDNA elution from Sepharose CL-4B in the presence of (NH

The second step was designed as a concentration step, and therefore, a support with high capacity for large molecules was used: a reverse phase POROS R1 50 matrix with a dynamic binding capacity between 5 and 1.5 mg pDNA/mL using a high flow rate (500 cm/h). The high flow rate is the main advantage of POROS. It does not affect the resolution or capacity during  pDNA purification because of the intraparticle, convective, solute transport, which is a fundamental new approach to reducing mass-transfer limitations in chromatography. This new approach will dramatically decrease separation time and increase throughput and productivity for pDNA recovery.

Another size-exclusion chromatography method was chosen as the last purification step. Agarose gel electrophoresis of column fractions showed that purity of the pDNA increased from the first fraction, which had <60%, to the last, in which purity of >95% was obtained (Figure 3). The pDNA fractions with a purity >89% were pooled and further concentrated by a 10% w/v PEG-8000 precipitation, which concentrated the plasmid and reduced levels of 

 host proteins and undesirable DNA contamination.

 The pellet was resuspended in the formulation buffer at 2 mg/mL and was finally filtered (0.22 μm) with a yield of about 100% before further analyses were performed. Analytical methods were followed according to the criteria recommended by the FDA. A summary of the analytical specifications and final results for pIDKE2 is shown in Table 1.

Table 1. The purified plasmid DNA had 95% purity. Contaminants such as host-cell RNA and DNA were undetectable by agarose gel electro-phoresis assay. Plasmid identity was confirmed by restriction enzyme digestion and  activity was confirmed by an in vivo assay, in which the results show that the purified pIDKE2 plasmid induced a positive and long-term antibody response against HCV core protein and enveloped proteins. 

The immunogenicity of pIDKE2 plasmid obtained by the method described in the current work was compared with another method previously described.

 All animals immunized with pIDKE2+Co.120 developed anti-HCV antibodies. Figure 4 shows the antibody response against HCV structural antigens at week 17. Mean antibody titers above 1:800 were elicited against HCV structural antigens in animals immunized with the mixture of pIDKE2 and Co.120. In fact, antibody titers against Co.120 and E2.680 reached values above 1:2,000. No significant differences were observed in antibody titers generated by immunization with the plas-mid obtained by different procedures. Mice immunized with normal saline did not induce any detectable antibody response. 

Historically, highly purified pDNA recovery has been accomplished through the use of cesium chloride/ethidium bromide (CsCL/EtBr) buoyant density gradient separation.

 This method allows the separation of pDNA by buoyant density into purified bands of different forms: supercoiled (sc), open circular (oc), linear (l) and multimeric (m) plasmid. Although it yields a highly purified plasmid, this approach is not scalable because of personnel safety issues and the hazardous waste considerations associated with the use of cesium chloride and ethidium bromide. Using these process solutions at large scales requires safety measures such as designing explosion-proof facilities or using appropriate protection. In addition, the use of ultracentrifugation is also a major impediment to the scale-up of this technology. In contrast, simple unit operations and the avoidance of critical reagents such as animal-derived compounds (e.g., enzymes), detergents, and organic solvents significantly reduce the need for validation efforts and precautions to ensure patient and operator safety.  

Our pDNA purification process is based on alkaline lysis, TFF, and size-exclusion chromatography as primary downstream steps for extensive removal of RNA. Reverse phase interaction chromatography is then used to purify the pDNA from the remaining impurities, particularly because of its ability to reduce the endotoxin burden to levels below the specifications. Volume reduction of the resulting stream is achieved by precipitation of the plasmid with PEG instead of 2-propanol or ethanol. Finally, size-exclusion chromatography is used as a polishing step and to exchange the buffer for an adequate formulation. The proposed process does not use or generate significant amounts of hazardous materials and no special safety requirements are envisaged. Thus, environmental or safety-associated costs are minimized. The reagents used do not pose any special regulatory concern because they are nontoxic, nonmutagenic, and nonflammable.  

It is also strongly recommended to spend sufficient time and efforts developing large-scale GMP processes. This may result in a different approach when compared with 'kit' protocols, in which convenience and simple robustness play the most important role. Depending on the final application as a therapeutic (high-dose single injection, or long-term, low-dose treatment) or for diagnostics, specific demands may require individual solutions. Given the complexity of the starting material, certainly single purification step will not be enough to meet the regulatory requirements. Nevertheless, the aim is to establish a robust and preferably generic protocol that is applicable to a variety of plasmids of different sizes (regardless of individual precautions related to stability or sensitivity to shear forces). When developing a multistep large-scale pDNA purification process, the design will aim to begin with fast volume reduction. This can be achieved by ultrafiltration or any (chromatographic) capture step, in which recovery (>90%) is more important than maximal capacity.

Currently published processes for pDNA purification include precipitation and extraction of pDNA by organic solvents, ultrafiltration, and predominantly liquid chromatographic techniques. Most of the available processes for pDNA purification are time-consuming and not scalable. Furthermore, because these processes use materials that are not certified for application in humans and also enzymes of avian or bovine origin, these processes do not meet the appropriate regulatory guidelines.  

Chromatography is considered the highest resolution method, and therefore, it is essential for producing pDNA suited for therapeutic applications. The most commonly used techniques for initial purification are anion-exchange and hydrophobic interaction.

 It has to be considered that the large pDNA molecules adsorb only at the outer surface of particulate supports.

 Consequently, capacities are usually on the order of hundreds of micrograms of plasmid per milliliter of chromatographic support. In our process, we used a POROS R1 50 reverse-phase matrix, which has a dynamic binding capacity between 5 and 1.5 mg pDNA per mL support. Finally, as a polishing step, size-exclusion chromatography is the most suitable for removing undesired pDNA isoforms and host cell proteins and to achieve buffer exchange. The processes recover 95% of pDNA similar to the process described by Horn, et al.

 The results demonstrate that this process meets all regulatory requirements and delivers pharmaceutical grade pDNA. The final chDNA content is <5 μg per dose, RNA is not detectable by agarose gel electrophoresis, protein content is lower than 5 μg per dose, and the endotoxin content is 0.6 EU per kg of body weight.  

The clinical application of gene therapy and DNA immunization will depend not only on efficacy but also on safety and the ease with which the technology may be adapted for large-scale pharmaceutical production. The  new purification process described here combines novel and conventional purification technologies as high-throughput alternatives. TFF membranes and a perfusion media were used for concentration and recovery of the pIDKE2 plasmid to reduce process time and to increase the productivity of the downstream process. Although the yields are lower when using this process, the advantages of this method over existing plasmid DNA purification methods include improved plasmid purity and the elimination of undesirable process additives such as toxic organic extractants and animal-derived enzymes. Finally, this process may be used as a simple, scalable, and applicable method for the production of the pIDKE2 plasmid, which is being used in Phase 2 human clinical trials. 

 is a technician in the downstream process development department, 

is a researcher in the analytical development department, 

is the head of the hepatitis C vaccine department, all at the Centre for Genetic Engineering and Biotechnology, Havana, Cuba, +53.7336.008, 

 1. Sandberg L, Bjurling A, Busson P, Vasi J, Lemmens R. Thiophilic interaction chromatography for supercoiled plasmid DNA purification. J Biotechnol. 2004;109:193–199. 

Features

Plasmid DNA Recovery Using Size-Exclusion and Perfusion Chromatography

Call it the awkward age: that critical period when a biotech company is no longer a start-up but is certainly not yet mature. As in late adolescence, when ill-defined dreams meet the specific demands of reality, the time comes when a company has to decide what it wants to be when it grows up.

It's not as easy as it might appear. Successfully coming through this critical period requires that the company: 

understands the strategic options and associated cost;

knows when it is time to make a choice; and

assesses the risks, timelines, commitments, and potential rewards of each option.

Selecting the right options requires careful thought, planning, and timing. A hasty decision can prematurely foreclose a more appropriate option. Deferred decisions can result in the business drifting in several directions at once, instead of proceeding directly toward a clear destination. In both cases the result is likely to be, at best, lost opportunity and, at worst, business failure. In our experience, it is far more profitable for up and coming innovative biotechs to take control of their destinies by making a careful strategic choice at the right time and then pursuing that option single-mindedly.

For the early-stage innovative biotech company, there are essentially two strategic options for the long term: (1) to become an idea factory or (2) to become a fully integrated company. Of course, there are ways to blend these two options, but the two basic options remain.

The idea factory is designed to generate innovative therapeutic products (with a strong IP position), devices, processes, or technologies and develop them to a point where they can be outlicensed to another company. The license then produces royalty payments that provide funding for generating new ideas which continue the cycle. In a variation of the strategy, idea-generating companies, especially those that have only one novel product or platform technology, may aim to be acquired by another company.

By contrast, the fully integrated firm takes ideas and innovations from discovery and development all the way to the market. Becoming such a company requires all of the functions and elements of infrastructure to execute the enormously difficult, complex, time-consuming, and costly job of commercializing a therapeutic product: R&D, quality assurance, quality control, regulatory affairs, clinical trials design and execution, technology scale-up, manufacturing, marketing, sales, and more. Further, the kind of executive management team required to operate a fully integrated company differs substantially from that required to run a successful idea factory.

There are, of course, other business models, such as a hybrid strategy that seeks to generate ideas, outlicense them, and use the royalties to commercialize other products. Firms that have unique platform technologies can profitably exist in this hybrid space as long as they have a strategy that limits their investment dollars to a manageable number of internal product candidates. But the idea factory and the fully integrated firm are the two proven business models that typically represent the dramatically different options biotechs must eventually confront.

Knowing when it is time to decide can be as important as the decision itself. In the beginning, the strategic decisions of many biotech companies are driven largely by two factors: by the belief of the founders in their scientific or technical expertise and by the investors who provide the early funding. Both factors are compelling and important, but they should not cloud the judgment of company decision-makers, either encouraging the postponement of hard strategic choices or propelling hasty, ill-considered decisions. In both cases, the result is likely to be an unrealistic or suboptimal strategy.

Innovative scientist-founders understandably have great faith in their ideas, which often represent exciting, if distant, novel possibilities for therapeutic breakthroughs. However, many of those ideas are initially born in talent-rich university laboratories, where considerations such as the highly regulated road to drug development, the demands of clinical trials and manufacturing, and the sheer scale of commercial operations are usually not well understood or practiced. Innovation is often the most highly coveted and rewarded skill set in these companies, and individuals are often promoted to high levels of executive management beyond their capability to run the practical day-to-day operations necessary to advance promising therapies to clinical trials.

Having evolved from that context and with little experience of the realities of drug development, many early-stage biotechs greatly underestimate the difficulty, the amount of time, and the costs of moving an idea from the pure research laboratory to a marketable product. At the same time, they often greatly overestimate the value of their idea—seeing it as potentially the most important development in its therapeutic area in decades—and they underestimate the long and arduous path to product approval. As a result, they may simply proceed on the unexamined assumption that full commercialization of the company's product or products is their goal, failing to recognize that successful business strategies usually require highly conscious choices.

Because capital is king in the very early stages of a company, it often seems unnecessary to think too deeply about ultimate strategy anyway. In fact, the earlier the stage of the company, the more likely it is that investors are making the decisions through active participation of the board of directors. And because financial angels and Series-A investors usually have very short investment horizons, most of those decisions will be based on an early exit strategy, not on longer-term prospects. But investors—especially after the Series-A round—also can have their judgment clouded. Tantalized by possibilities of the greatest possible return on the investment, they, too, may defer making a final decision on the course the company should take. Conversely, unduly alarmed by early setbacks in research or development, they may make decisions that should in fact be deferred—and prematurely deciding the company's future can sometimes be as detrimental as putting off such decisions.

How do you know when it's time to decide? What information is necessary to make serious additional investments in the company and its ideas? There is no hard and fast rule, but there are a number of factors that point to the window of opportunity.

First, there is the subjective feeling that comes after you have survived the early rounds of funding, when, as a decision-maker, you realize that you no longer have to think almost exclusively about capital. You've made it into subsequent rounds of funding with investors who may have longer investment horizons and your product is showing real promise.  

Second, there are some objective milestones that you should have achieved. These include:   

a platform technology or drug candidate with good prospects of success;

well-established and defendable intellectual property coverage;

comprehensive characterization of the product, including stability;

thorough and positive preclinical and toxicology results;

a robust, reproducible, and cost-effective production process; 

indications of the product's manufacturability;

reasonable assurance that the product is innovative;

a positive assessment of the reimbursement landscape for the projected therapy; and

early and encouraging meetings with the FDA.

Third, regardless of which strategy is chosen, many companies wish to reach the point of having Phase 1 and Phase 2 clinical trial data to support claims of clinical safety and efficacy, respectively. If so, the company will need to be prepared to:  

compile the appropriate documentation to support investigational new drug (IND) filings;

decide what to do internally versus contract out;

be prepared to address consequences and show comparability as required.

ASSESSING RISKS, REWARDS, AND REQUIREMENTS 

Because the differences in risks and rewards between the idea factory and the fully integrated firm are so vast, it is essential to make a well-timed, unambiguous decision that will guide implementation of the strategy and sharply focus the efforts of the company on the ultimate goal. Those efforts include getting the type of management team in place that is appropriate for the strategy. For the idea factory, that usually means a small team that is adept at research, early development, alliances, and outlicensing. For the fully integrated firm, it means strong executive leadership in critical commercial areas like manufacturing, marketing, and sales. (For a discussion of how business processes and infrastructure should be created and measured for each strategic alternative, see 

 March 2008 Compliance Notes column, "How to Avoid Becoming a Biotech Zombie—Part 3".)

The risks for the idea factory are far lower than the risks for the fully integrated firm. The products that the idea factory produces need only show some therapeutic promise. They do not have to be herded all the way through regulatory approval and use in humans.

Not surprisingly, lower risk means lower rewards. Nevertheless, for the idea factory that can generate a continual stream of ideas and then turn them over to another company for commercialization, those rewards can be reasonably consistent and substantial. Moreover, with big pharmaceutical companies facing thin pipelines and patent expirations, the market for ideas is strong.

The fully integrated company stands to reap even more substantial rewards, but the risks are concomitantly far greater than the risks for the idea factory. The fully integrated firm must make enormous investments in human and infrastructure resources. Further, for every drug that doesn't make it to full commercialization or that fails to win sufficient market share, those investments are wasted. And the odds of successful commercialization are long. For example, it is estimated that an IND has a 10–15% chance of reaching the market, and an alarming number of new drugs (50% by FDA estimates) are failing trials in late stages. The fully integrated company must therefore have enough promising products in the laboratory and the clinic as a hedge against products that fail.

Deciding between these two risk profiles is not a simple matter. Certainly, different companies, their boards, and their executives will have different appetites for risk, often depending on who among those stakeholders is ultimately making the decision. But regardless of who makes the decision, it should be made at a much greater level of granularity than a simple macrochoice between low risk or low return and high risk or high return.  

With an understanding of the differing requirements for the two business models, you can clearly and comprehensively define your risks, both external and internal: competitor actions, patent protection, gaps in core competencies and management, product development, regulatory hurdles, and other risks that could delay reaching the ultimate goal under each model. You can then consider what potential effect these risks are likely to have on cash flow and model the cash flow under different scenarios.

Unfortunately, many organizations treat risk and uncertainty in a highly unsystematic fashion. They may have a general, qualitative discussion of potential setbacks and pitfalls, but they fail to try to quantitatively calculate the chances that particular problems will occur. Instead, they simply increase the discount rate to compensate for a vaguely conceived notion of risk. But this fails to account for the specific risks of a chosen strategy. Those specific risks must be quantified in terms of their contribution to uncertainty in cash flow. Although most biotech companies lack the ability to conduct this kind of probabilistic analysis, it is nevertheless worthwhile to seek assistance with it to understand the total risk exposure of any given business plan. Then it is possible to make a fully informed decision about which business model to pursue.

Regardless of which path you choose, you will then face another crucial set of decisions that turn on risks and rewards—whether to build or buy such capabilities as:  

quality control laboratory testing capability for drug release and stability testing;

internal versus CRO clinical study capabilities; and

It is this build-or-buy analysis that is the subject of the next article in this series.

 is a principal, all at Tunnell Consulting, King of Prussia, PA, 610.337.0820, 

Every biotech company reaches a point in its development where it must decide what path it will take after it passes the start-up phase. This article discusses what the company must consider to decide what business model it will follow.

The Company at the Crossroads. Part 1: To Commercialize or Not?

Removing viral contaminants from animal cell-culture derived biologicals is a major challenge of downstream purification because it involves laborious and time-consuming techniques that result in increased manufacturing costs. Updated regulatory guidelines demanding higher safety margins and enforcing good manufacturing practices are leading to tighter specifications. This stresses the need to implement robust and efficient orthogonal strategies for virus clearance to meet the requirements of a virus-clearance approach based on risk assessment. Such technologies can involve virus removal by nanofiltration, inactivation by ultraviolet C (UVC), and adsorption by membrane chromatography. Additionally, this three-tier platform should be characterized by using disposables to meet the flexibility and low capital requirements needed in early-stage process development. All of these new paradigms in virus clearance are scalable, economical, orthogonal, and disposable.

Today's downstream processing operations generally focus on two main areas: the initial recovery phase, when bulk purity is achieved, and the subsequent polishing phase, which adds safety through orthogonal strategies for impurity and pathogen clearance.

  Commercial manufacturing of therapeutic antibodies requires robust and reliable processes that are economical and deliver high yields of a product that is pure and safe for human use. One factor that poses a constant threat to product safety is the presence of viruses in the finished product. Virus contamination of products derived from human or animal cells can have disastrous clinical consequences causing diseases ranging from common colds and influenza, to acquired immune deficiency syndrome (AIDS), hepatitis, herpes, measles, and poliomyelitis. Some viruses like Epstein-Barr, human papillomavirus, and retroviruses are even oncogenic, causing the insertion of cancer-causing genes into cellular genomes.

  It is essential to review both the short-and long-term consequences of viral contaminants existing in biopharmaceutical products. In this context, it is worthwhile to understand some important aspects of current and state-of-the-art methods for inactivating and eliminating viruses from process streams that generate products intended for use by humans.  

Viruses are composed of small amounts of DNA or RNA, encapsulated by a protein coat, and may be enclosed in an envelope made of proteins, carbohydrates, and lipids. Viruses exploit the enzymes and other host-cell machinery to replicate themselves. The viral nucleic acid can be single-or double-strand DNA or RNA. A single virion is a completely developed virus particle made of 1–50% nucleic acid and 50–99% proteins or glycoproteins and lipids. Virions range from about 15 to 450 nm in size.  

REGULATORY REQUIREMENTS REGARDING VIRAL INACTIVATION AND CLEARANCE 

Viral contamination is a risk to all biotechnology products derived from cell lines of human or animal origin. Contamination of a product with endogenous viruses from cell banks, or adventitious viruses from personnel can have serious clinical implications.

  To ensure maximum viral safety, the ICH Q5A regulatory guideline mandates that manufacturers of therapeutic biological products for human use implement adequate technologies in their manufacturing process and demonstrate the capability of their processes to remove or inactivate known or adventitious contaminants based on a process-specific virus clearance strategy.

  According to the latest draft on regulatory guidance from the European Agency for Evaluation of Medicinal Products (EMEA), potential contaminants may be enveloped or nonenveloped, small or large, DNA or RNA, labile or resistant viruses.

Viral safety of licensed biological products must be assured by three complementary approaches: (i) thorough testing of the cell line and all raw materials for viral contaminants, (ii) assessing the capacity of downstream processing to clear infectious viruses, and (iii) testing the product at appropriate steps for contaminating viruses.

  The first study is required before Phase 1 clinical trials, in which the process should be evaluated for inactivation or removal of an enveloped and a small nonenveloped virus and at least two orthogonal steps should be used for achieving the same.

  A second and more complex study is then conducted before manufacturing Phase 2/3 materials to provide evidence of the effective and adequate clearance of relevant and known viruses, as well as the removal of a range of novel and unpredictable viruses.

  A viral clearance study with at least four viruses for late stage is state-of-the-art and lower-range values (LRV) of four or higher are perceived as robust and effective safety measures.

  At least one of these clearance steps evaluated in a validation study must be effective against nonenveloped viruses, such as porcine parvovirus (PPV), canine parvovirus (CPV), or minute virus of mice (MVM). As for enveloped retroviruses, although no cases of infection or transmission of Chinese hamster ovary (CHO) cell-related type A and C virus particles have been reported so far, retrovirus-like particles theoretically pose safety concerns to humans because of their morphological and biochemical resemblance to tumorigenic retroviruses, and therefore, have to be completely removed or inactivated.

METHODS FOR VIRAL INACTIVATION AND CLEARANCE

The robust and reliable capability to eliminate viruses must be demonstrated by a risk-based approach.

  Today's requirements demand a statistically independent combination of methods (orthogonal technologies) for removing enveloped and nonenveloped viruses based on the different physical principles of removal and inactivation, and yet are complementary to each other.

Several methods can be used for virus clearance in bioprocessing. These include inactivation methods such as solvent and detergent (SD) or chemical treatments, low pH, microwave heating, adsorption by chromatography, and removal by mechanical or molecular sieving using normal and tangential-flow filtration methods. The first three of these, treatments with solvents and detergents, low pH, or microwave heating, all have significant limitations in their ability to inactivate small nonenveloped viruses. SD treatments were commonly used for plasma proteins and were considered the gold standard for inactivating enveloped viruses.

  It has been shown that SD treatments of a recombinant protein can completely and rapidly inactivate enveloped viruses like PI-3, XMuLV, IBR, and MCF.

  However, small nonenveloped viruses are not being eliminated substantially by this virus-clearance technology. Low pH inactivation of murine retroviruses is reported to be highly dependent on time, temperature, pH, and relatively independent of the recombinant protein type or conductivity conditions outlined.

  Heating is not considered one of the most reliable methods for virus inactivation because of the variation in stability of each viral genome to heat or temperature.

VIRAL CLEARANCE CAPABILITIES OF MEMBRANE CHROMATOGRAPHY 

Chromatography and filtration, on the other hand, are widely accepted methods for virus adsorption and removal respectively and act as orthogonal techniques in the viral-clearance platform. Membrane chromatography, a relatively newer technique gaining prominence in biomanufacturing, has proven to be efficient in removing small nonenveloped viruses.

  Ion-exchange membrane adsorbers, with ligand–virus-binding properties similar to those of anion-exchange (AEX) chromatography, have the disposable option as an added advantage. This not only reduces capital costs but also eliminates post-use cleaning, sterilization, validation, and risk of carry-over contamination, thereby simplifying adsorptive virus clearance.

Efficient clearance data between 4.41 log

 for MVM has been determined for membrane chromatography.

  Additional studies have demonstrated that membrane chromatography meets and exceeds viral-clearance performance of Q resin chromatography.

  Clearance capabilities of Sartobind Q for nonenveloped viruses have been shown to be between 3.56 log

  It has been demonstrated that the platform tested membrane chromatography, has a process capacity greater than 3,000 g MAb/m

  or 10.7 kg MAb/L with a LRV >5 for four model viruses.

  Mass balance in viral-clearance study is another important parameter to demonstrate efficient virus removal by membrane adsorbers. A 100% recovery was demonstrated for PRV, Reo-3, and MVM, when the membrane was stripped with 1-M NaCl, illustrating efficient charge capture for the three model viruses while high salt treatment of the membrane showed 70% recovery for MuLV.

  The virus-clearance capability of such technology has been presented in Tables 1a and 1b. 

Table 1a. Process capacity and virus adsorption capability of Sartobind Q membrane adsorber16 

Table 1b.  Lower-range values (LRV) of different viruses demonstrating mass balance upon adsorption on Sartobind Q17 

A second orthogonal technique in the virus-clearance platform is nanofiltration that has traditionally been accepted as a robust method for virus clearance.

 This is the most expensive downstream step, accounting for up to 40% of costs, and is the natural target for optimization.

 Initially, virus removal by filtration was found to be highly dependent on size of the virus, and less dependent on parameters like buffer composition, process time, protein type, and pressure.

 Earlier studies have shown the principal feasibility of PP7, a small nonenveloped 25 nm bacteriophage, to act as a model virus for small, nonenveloped viruses.

  The ability of commonly available nanofilters to retain bacteriophage has been clearly demonstrated in recent studies.

  Virus spiking trials using 20-nm retentive virus removal filters have also shown to clear both large and small viruses (Tables 2a and 2b).

Table 2a. Virus retention capacity of Virosart CPV as demonstrated by TCID50 assay26 

Table 2b.  Virus retention capacity of Virosart CPV as demonstrated by Plaque assay27 

The evaluation of a virus filter should not be limited only to its capacity. An ideal virus filter should retain all viruses and allow high protein transmission while maintaining a high flow rate without significant virus breakthrough. Unexpected virus passage during a virus-clearance step is undesirable from a good manufacturing practice (GMP) and validation standpoint.

 Virus passage through the filter could occur for many reasons, including accumulation of aggregates, high spike concentrations, and other impurities that may block the filter or result in a breakthrough.

 This is a serious safety concern that must be minimized. Although contaminants and other various parameters may be the main cause of filter breakdown, some nanofilters still efficiently remove viruses at high LRVs even when experiencing high flow decay. Earlier, detailed analysis of the retention characteristics of PP7 by a PESU-based 20-nm nanofilter underlined the principal capability of nanofiltration to act as a robust and effective virus removal step independent of flow decay or the nature of product being filtered (Figure 1).

  Additionally, a recent report outlined the various titer reduction capabilities of virus retentive nanofilters.

  The study showed that not all filters tested for their LRVs versus flow decay profile experienced a significant loss of titer reduction with increasing flow decay. To ensure the highest level of viral safety of biopharmaceuticals, it is important to understand and predict the efficiency of virus removal steps while also realizing that small virus-retentive filters should not be viewed as absolute in their capacity to clear viruses. 

In this article, we propose implementing ultraviolet C (UVC) as a third orthogonal technology for virus clearance in the downstream purification process. UVC, newly developed as a virus-inactivation technology, targets small, nonenveloped viruses and offers another robust method for removing adventitious viruses. In UVC radiation, low-dose radiation at 254 nm destroys the viral nucleic acid while maintaining the structural and functional integrity of the protein of interest. IgG losses are <5%.

  The efficiency of viral inactivation and product recovery is sensitive to the viscosity and absorption coefficient of the protein solution and its residence time in the radiation chamber.

How to implement a risk-based approach to eliminating viruses.

BioPharm International-09-01-2008

Protecting Patents for Personalized Medicine

Minimizing the Problem of OOS

Plasmid DNA Recovery Using Size-Exclusion and Perfusion Chromatography

The Company at the Crossroads. Part 1: To Commercialize or Not?

Viral Clearance Strategy Using a Three-Tier Orthogonal Technology Platform

Global Manufacturing Challenges Industry Regulators

The Art of the Deal

The Outlook for Quality by Design