What is the safety ratio of a drug?

Pivotal Margin of Safety Study

Margin of safety studies have historically been used to support the safety of an investigational new animal drug. These studies are generally characterized by a small sample size, relative homogeneity of study animals, limited study duration, and the use of healthy young animals. Although the use of multiple doses is commonly needed in order to extrapolate safety findings of new animal drugs to their use under various clinical conditions, the actual multiples of the 1X dose in a margin of safety study are not strictly defined. Most typically, however, the margin of safety is demonstrated in a 0X-, 1X-, 3X-, and 5X-dose study with the drug administered for 3X the intended duration. The product safety is then established by demonstrating an acceptable level of safety (above 1X dose) and identifying (if present) the toxic syndrome. Variables that are typically assessed in a margin of safety study include physical examinations and observations; various clinical pathology tests (hematology, blood chemistry, and urinalysis); necropsy; and histopathology. Other information, such as toxicokinetic data, may also be collected if deemed necessary.

Acute Exposure

The margin of safety for each population subgroup for theoretical acute dietary exposure to cyanazine is given in Table 33.8. These values were derived from the theoretical dietary exposure values (Table 33.6) in which all registered commodities were assumed to contain residues at the default level of the LOD. The MOS values ranged from 6270, for nonnursing infants (< 1 year), to 26,300 for seniors (55 + years). The inclusion of theoretical drinking water residues at 0.1 ppb (LOD) reduced the MOS for the U.S. population, all seasons, from 13,500 to 12,800, a 5% decrease.

Table 33.8. Margins of Safety for Theoretical Dietary Exposure to Cyanazine Residues in All Commodities with U.S. EPA Tolerancesa

5.8 Frequently Used Terms for Safety Evaluation of Drugs

The margin of safety of a compound is determined from the results of two studies, such as an ED study and an LD study. A ratio between selected ED and LD values is then used to express the margin of safety. The larger the ratio, the greater is the margin of safety.

TI may be defined as the ratio of the LD50 and the ED50.

TI=LD 50/ED50

where LD50 is the dose that is lethal for 50% of the population and ED50 is the dose that is effective for 50% of the population.

The TI measure is commonly used for evaluating the safety and usefulness of therapeutic agents. The higher the index, the safer is the drug.

The TR may be defined as the ratio of the lethal dose-25 (LD25) and the effective dose-75 (ED75).

TR=LD25/ED75

where LD25 is the dose that is lethal for 25% of the population and ED75 is the dose that is effective for 75% of the population.

TR is considered a better index of safety of a compound because it also includes the steepness of curve. In toxicity cases, a flatter curve is considered more toxic, or hyper-reactive groups are at a much greater risk than the hyporeactive or normal group. Shallower curves usually have low therapeutic ratios.

SEM may be defined as the ratio of the LD1 and the effective dose-99 (ED99).

SEM=LD1/ED99

where LD1 is the dose that is lethal for 1% of the population and ED99 is the dose that is effective for 99% of the population.

The SEM is a more conservative estimate than the TI because values are derived from extremes of the respective dose–response curves.

Chronicity factor is the ratio of the acute LD50 (one dose) to chronic LD50 doses.

Chronicityfactor=acuteLD50/chronicLD50

The chronicity factor is used to assess the cumulative action of a toxicant. Compounds with cumulative effects have a higher chronicity factor.

The ratio between the inherent toxicity and the exposure level gives the risk ratio. Risk ratio indicates the risk of a compound. Substances of higher inherent toxicity may pose little risk because access of exposure of individuals to such agents is limited. Compounds of low toxicity may be dangerous if used extensively.

Drinking Water Ingestion

The LADD (mg/kg/day) from drinking water ingestion for an individual is calculated using Equation (31.1):

(31.1)LADD=[concentration of herbicide in drinking water, mg/liter(L)]×(amount of drinking water ingested per day,L/d)/(body weight in kg)

Probability distributions for the LADD in the 18 states that use the most atrazine and simazine every year (approximately 90% of the total) were determined. In these major-use states, the concentration of herbicide in the drinking water varies too much between and within states to be accurately characterized by a single number. Instead, the database of observed individual concentrations collected by the states for local community water supplies, and the number of people served by each community water supply, were used in the Monte Carlo evaluations of Equation (31.1) and the corresponding LADD distributions. In determining these LADD distributions, the objective is to make the person whose drinking water herbicide concentration is used in Equation (31.1) equally likely to be each person served by community water supplies. For example, if the population of interest is a state, then the LADD distribution in that state is determined by randomly selecting a large number of individuals from that state and randomly selecting each individual's drinking water concentration from the database of drinking water concentrations for that individual's community water supply system. In order for the resulting distribution to correspond to the state's distribution, the selection process is done in a way that makes each person in the state equally likely to be selected and makes the likelihood of a community water supply being selected equal to its relative size within the state (i.e., the number of individuals served by the community water supply system divided by the number of individuals in the state). If the population of interest includes more than one state, then individuals are selected so that each individual in the population is equally likely to be selected, and the likelihood of each state is proportional to the relative size of the state within the total population.

Because the variability in the amount of drinking water ingested per day per kilogram of body weight is much smaller than the variability of the atrazine and simazine concentrations in the drinking water, Equation (31.1) is evaluated assuming a default upper bound value of 2 L/d and a default adult body weight of 70 kg/d.

The distributional analysis of the dose from exposure using Equation (31.1) indicates that for atrazine, at least 95% of the estimated LADDs from drinking water ingestion have an MOE of at least 50000 in the 18 major-use states combined for atrazine (Figure 31.1). Figure 31.1 shows a histogram of the MOEs for atrazine in the 18 major-use states combined. The horizontal axis indicates intervals of possible MOE values, and the vertical axis indicates the proportion of individuals in the 18 major-use states that are estimated to have MOEs in that interval. For example, the smallest MOEs in the population are in the interval from 1000 to 5000; the proportion in this interval is only 0.0013 (0.13% of the population). The proportion of the population with MOEs below 50000 is approximately 0.05 (0.0013 + 0.0065 + 0.0443 =0.0521). Hence, approximately 95% of the MOEs in the population are greater than 50000. Figure 31.1 indicates not only the 95% lower bound on the MOE, but also the entire MOE distribution. This distribution covers a range from 1000 to more than 10 billion, which indicates that the MOE in the population is quite variable and that most have MOEs considerably above the 95% lower bound. For simazine, at least 95% of the MOEs are greater than 200000 in the 18 major-use states combined.

Figures 31.2 and 31.3 show the atrazine MOEs for the 18 individual major-use states separately. The entire histograms in these figures are not all easily seen, but what is important is that these major-use states have hardly any MOEs below 5000 and that most of the people in every state have much larger MOEs.

The margins of safety indicated by the MOEs in Figures 31.1–31.3 are even greater when the exposure evaluation is expanded to include the following alternatives:

Drinking water consumption distribution and body weight distribution.

Age-dependent drinking water consumption and body weight distributions.

Year-to-year variability as opposed to the same concentration and consumption for 70 years.

Exposure duration distributions corresponding to residential durations as opposed to 70 years.

More recent water monitoring data. (Data used in this assessment were collected prior to June 1994, and atrazine levels in water are declining.)

Dietary Consumption

The LADD (mg/kg/day) from dietary exposure can be calculated for an individual in a specified population or subpopulation, using Equation (31.2):

(31.2)LADD = sum of the dose from each food

i = number of foods

= ∑i=1 [(amount of foodi consumed in a day per kilogram body weight, mg/kg/d) × (residue concentration in raw agricultural commodity contributing to foodi, mg herbicide/mg food) × (adjustment factor 1 for foodi) × (adjustment factor 2 for foodi)]

In Equation (31.2), the amount of each type of food consumed in a day per unit body weight of the consumer is assumed to be a constant, equal to the corresponding food consumption value in the USEPA's database for Dietary Risk Exposure Assessments (DRES), which is an average chronic consumption value [US Department of Agriculture (USDA), 1983]. For most consumed foods, the food originates as a raw agricultural commodity. The fraction of the weight of the raw agricultural commodity that is the chemical of interest (e.g., atrazine or simazine) is the residue concentration. The residue concentration in the raw agricultural commodity is not necessarily the same as the chemical's concentration in the food as it is consumed. For example, the concentration of a chemical in an ear of corn when it is harvested in the field and the concentration after it has been cleaned and cooked may be different. This difference is accounted for by ‘adjustment factor 1.’ The values for adjustment factor 1 are the default constant values in DRES. During an individual's lifetime, some of the raw agricultural commodity in consumed food may come from crops treated with the chemical of interest, and some may come from untreated crops. An individual's lifetime average proportion from treated crops is assumed to equal the proportion of acres treated with the chemical. This proportion is reflected in Equation (31.2) as adjustment factor 2. The constant values for adjustment factor 2 were the data available on percent crop acreage treated. Ciba Crop Protection obtained these data in 1993 from Maritz Marketing Research Inc. and from Doane Marketing Research Inc., both of St. Louis, Missouri. In sensitivity analyses, adjustment factor 2 can be set at 1.0 to correspond to an individual's food being all locally produced and treated, instead of having residue concentrations corresponding to the national average.

Macadamia nuts, guava, refined sugar, and molasses are the only raw agricultural commodities treated with atrazine that are consumed as foods. There are no known residue concentrations of atrazine or its chloro-metabolites above their analytical limits of detection (LODs) in any of these four foods. In evaluating Equation (31.2), the residue concentration in each of these four foods is assumed to be equally likely to be any value between zero and its LOD (i.e., uniformly distributed between zero and the LOD).

For meat, milk, and eggs, the residue concentration in raw agricultural commodity contributing to foodi in Equation (31.2) is the concentration of the chemical of interest that results from some of the raw agricultural commodities in the diets of cattle and poultry being treated with that chemical. While the observed residue concentrations in meat, milk, and eggs are below their LODs, the concentrations of atrazine in the raw agricultural commodities used as feed for cows and poultry are sometimes quantifiable. Probability distributions on the anticipated residue concentrations of atrazine and its chloro-metabolites in meat, milk, and eggs are based on estimated diets for cows and poultry, the observed residue concentration distributions in the components of these diets, and proportionality constants relating high experimental concentrations in feed to resulting concentrations in meat, milk, and eggs (Sielken et al., 1996). The estimated diets provided adequate nutrition to poultry and lactating dairy cattle and maximized the amount of feed items treated with atrazine.

Using Equation (31.2), distributional analyses of dietary exposure to atrazine and its chloro-metabolites in the United States and the four regions (Northwest, North Central, Southern, or Western) indicate that at least 95% of the estimated LADDs from dietary consumption have an MOE of at least 300000 in each of the four regions and 330000 in the United States as a whole (Figure 31.4).

Figure 31.4. Distributions of the MOEs for atrazine plus its chloro-metabolites from dietary consumption.

The atrazine chloro-metabolites in the diet have been combined with atrazine in Figure 31.4. Atrazine's chloro-metabolites in the diet have been assumed to have the same toxicity as atrazine in calculating the MOEs in Figure 31.4.

There is not much variability in the distributions in Figure 31.4 because all of the components in Equation (31.2) have been assumed to be their lifetime average values, except for the residue concentrations. Thus, the only really important characteristic of the distributions in Figure 31.4 is that the MOEs are quite large.

For simazine, the residue concentrations in Equation (31.2) are constants (averages or upper bounds), determined directly from the most recent residue data on the commodities themselves or, for meat, milk, and eggs, determined indirectly from the diets of cattle and poultry. The corresponding MOE is at least 1750000 for each of the four regions and at least 2000000 for the United States as a whole.

While the following two observations are not critical in the distributional characterization of the intake of atrazine and simazine from dietary consumption, these observations can be important in other situations. First, making the assumption that the residue concentration in an individual's food is the same every time that food is consumed (as in Equation (31.2) exaggerates the variability in the intake distribution. Without this assumption, both the low and high percentiles of the intake distribution would be closer to the median intake, and the 95% lower bound on the MOE would increase. Second, when a sum is being characterized (such as the sum of intakes in Equation (31.2), it is important to determine explicitly the probability distribution of the entire sum and not to attempt to infer the characteristics of the distribution of the sum indirectly from the distributions of its components. For example, the 95th percentile of a sum may be much smaller than the sum of the 95th percentiles of its components.

Monitoring

The margin of safety during neurosurgical anesthesia can be improved by using specialized monitoring that provides information on brain function and circulation. Early detection of threats to brain function and circulation is important. Changes in anesthetic and/or surgical approach may alter the neurologic outcome of the patient.

Neurophysiologic monitoring such as electroencephalography (EEG) and evoked potentials (somatosensory evoked potentials [SSEPs], brain stem auditory evoked response [BAER]) are valuable tools for detecting threat to the brain and spinal cord. These clinical applications are discussed in another chapter. Computed tomography (CT) and magnetic resonance imaging (MRI) are discussed in the neuroradiologic section of this book.

Specific monitoring techniques are available that will improve the safety of patients who are placed in positions at increased risk for venous air embolism. Transesophageal echocardiography is the most sensitive device for detecting air in the heart (0.02 mL/kg), and it is the only monitor that can detect air bilaterally. Its disadvantages include the risk of recurrent laryngeal nerve damage from the TEE probe that can occur in the flexed neck of a neurosurgical patient. One also needs a trained observer to interpret the TEE data.

The precordial Doppler ultrasound transducer is a very sensitive device for the detection of air in the right atrium. Small amounts of air (0.2 mL/kg) are detected easily because air is a good acoustic reflector. It is recommended that this device be used in conjunction with a capnogram for the detection of venous air embolism. The arterial-to-end tidal CO2 gradient is increased with an air embolism. Low cardiac output and hypothermia may also alter this relationship.

Introduction

Over the last 15 years, both Raman spectroscopy (RS) and Raman imaging (RI) have gained an increasing attention in the field of biomedical diagnostics. RS and RI have been applied in both cell cultures and tissue studies by measuring the concentration of the bioactive substances in body fluids in in vivo, in vitro, and ex vivo.

RS and RI are increasingly used for diagnosis of many diseases. As the powerful potential of the methods has been realized, they are quickly making their ways into clinics. However, a number of challenges remain to overcome, especially in vivo diagnosis. In this article, the recent progress of RS and RI towards clinical diagnosis is discussed from a critical perspective. During the last decade, RS and RI have been applied to medical applications. This article reviews new trends of Raman applications in medicine. We focus on the applications of RS and RI medicinal development within the period from 2000 to now. Raman applications comprise a broad range of medical areas such as ophthalmology, oncology, pharmacy, dentistry, dermatology, gastroenterology, in medical research cardiology, orthopedics, neurology, urology, and reproductive medicine.

RS found a broad range of applications in ophthalmology, including chemical and structural characterization of pathologies, particularly extended cataracts studies, retina diseases, and keratitis.

In the field of dermatology, RS and RI have been used not only in detection of cancerous skin tissue, but also in determining skin morphology, the level of water content in the skin layers, their composition (for instance carotenoids concentration), and the permeability for biologically active substances such as retinol, phospholipids and metronidazole, hyaluronic acids.

Arteriosclerotic vascular disease studies have also been studied by RS in vivo and ex vivo. In human coronary artery tissue lipids pools, cholesterol and calcifications are distinguished, which is typical for atherosclerotic lesions. Another studies claim that proteins and lipid moieties are atherosclerotic plague indicators in carotid artery tissue.

Estimates of cancer incidence in the EU (worldwide) in 2012 indicate about 2.6 million new cancer cases with 1.2 million cancer deaths, thus emphasizing the role of cancer treatment as a central target in the medical research. Since the late 1990s, only incremental advances have been achieved. This slowing down in the treatment of cancer, together with the fact that some types of cancer are essentially insensitive to radio/chemo therapies (as, e.g., lung or brain) and the harmful whole-body secondary effects of those treatments, emphasize the urgent need to develop different approaches to cancer diagnostics, treatment, and monitoring responses to treatment. Current diagnostic and imaging methods in the clinical sector are expensive, sophisticated, time consuming, and are often limited by inadequate sensitivity, specificity, and spatial resolution.

Cancer diagnosis requires better screening of early stages of pathology and monitoring patient responses to treatment. RS and RI develop a powerful alternative for pre-operative and intra-operative cancer diagnosis, therapy monitoring, and drugs localization, which may bring revolution in cancer detection and treatment. The RI combined with atomic force microscopy (AFM), near-field scanning optical microscopy (SNOM), tip enhanced Raman spectroscopy (TERS) tools will “upgrade” cancer epigenetics and metabolic research and put together an ambitious plan to tackle many unanswered questions by monitoring the biochemistry/morphology/nanomechanics/dynamics of cells. Raman approach will provide unique insight into vibrational features of cells and intracellular processes occurring in normal and cancerous human tissues as well as localization of therapy drugs. This goal will be achieved by unsurpassed spatial resolution (down 10 nm), sensitivity (up to 10− 14 M), and specificity offered by the Raman approach.

The most significant application concerns the area of cancer diagnostics of breast, brain, colon, lungs, and skin. RS and RI are promising techniques for distinguishing the healthy and cancerous tissue and determining the cancer stage, invasiveness, and aggressiveness.

Breast cancer is a type of malignancy that affects breast tissue and its incidences are increasing across the globe. It is also one of the most common cancers in women which cause high morbidity and mortality rates. Recently, we have shown that looking inside human breast ducts can help answer fundamental questions about location and distribution of various biochemical components inside the lumen, epithelial cells of the duct, and the stroma around the duct during cancer development.

Vibrational signatures of epithelial breast cancer cell lines and human breast tissues (ductal and lobular carcinoma) have been used to identify and discriminate structures in normal and cancerous tissues. Our approach has reached a clinically relevant level in regard to cancer diagnosis. The proposed Raman biomarkers will allow for guidance of intra-operative tumor resection in real surgery time, and accurately delineate tumor margins.

The assessment of safety margin of breast cancer is very important in clinical practice during partial mastectomy as a surgical guidance tool. When a tumor is removed, some tissue surrounding is also removed. The safety margin, also known as “margins of resection,” is an area within the distance between a tumor and the edge of the surrounding tissue that is removed along with it in the surgery. A pathologist checks the tissue under a microscope to see if the margins are free of cancer cells. Depending upon what the pathologist sees, the margins of a tumor can be classified as

Positive margins: cancer cells extend out to the edge of the tissue.

Negative margins: no cancer cells are found.

Close margins: any situation that falls between positive and negative is considered “close.”

Figs. 1 and 2 show the typical Raman imaging spectra of the breast tissue surrounding the tumor from the safety margin and the cancerous breast tissue (infiltrating ductal cancer) from the tumor mass of the same patient.

In case of human breast cancer, the research concerns not only the cancerous (lobular and ductal carcinoma) and margin human breast tissue but also the cell lines such as M1, MR1, MCF10A, MCF7, MDA-MB-231, MCF10AT, MDA MB435, and Hs578T. Carotenoids, fatty acids, and their metabolic derivatives are suggested as the Raman biomarkers to this type of cancer, as well as the protein and genetic material to lipid content ratio is claimed to have an influence on diagnosis.

Extended studies on brain tumors such as glioblastoma, meningioma, gliomas, and metastatic tumor differentiation from normal brain tissue are possible by determining the lipids and proteins ratio and segregate the grade of development base on that. Gliobastoma studies were also carried out, concentration on distinguishing gray matter, necrosis, and cancerous tissue. Skin cancer can be diagnosed based on the lipid and protein profile in normal and cancerous skin and there is a possibility to distinct various types of cancers such as cutaneous melanoma pigmented nevi, seborrheic keratosis, and basal cell carcinoma. Cervical cancer cells were examined in human keratinocytes cell line (normal primary human keratinocytes (PHKs) and with Human Papilloma Virus HPV 16 E7 gene induced expression and CaSki cells) to discriminate them using Principal Component Analysis (PCA). Gastric adenocarcinoma is observed by the changes in protein and lipid profile of the normal and tumor tissue, which was confirmed by the correlated SGC7901 cell line and gastric mucosa tissue studies. For gastric cancer diagnosis, biomarkers were found as phospholipids and protein profiles in normal and cancerous tissue. Upper gastrointestinal tissue can be classified as cancerous or normal based on actin, collagen, triolein, and glycogen content. Adenocarcinoma cancer tissue and normal colonic tissue can be differentiated by the proteins and nucleic acid profile. Surgically resected colectomy specimens can be classified as normal or cancerous based on protein, DNA, and lipids profile. Lung cancer tissue is characterized by changed protein profile, especially in increased tryptophan and phenylalanine percentage and decreased phospholipids, proline, and tyrosine percentage in comparison with normal bronchial tissue. Lung cell line studies (normal lung cell line MRC-5 and cancerous RERF-LC-MS, EBC-1, Lu-65, RERF-LC-MA) indicated cytochrome c as a biomarker to distinguish normal and cancer cells. In cervical cancer studies, a possibility of specific detection of the cervical dysplasia by RS was found, which is not affected by physiological factors and patient's medical history. Leukemia studies claim that the DNA to protein ratio is important to differentiate T- and B-leukemia cells from normal Peripheral blood mononuclear cells (PBMCs). Cancerous bladder tissue and normal one can be not only distinguished but also classify the grade and invasive or non-invasive stage in vitro based on PCA.

The most lethal form of malaria, Plasmodium falciparum, can be diagnosed by RS, by determining the heme species in erythrocytes and antimalarial drugs interactions.

As we know stem cells provide unique “native” human cell types that give researchers new opportunities to study targets and pathways relevant to disease processes, to identify and confirm new drug targets.

RS and RI arm scientists with the tools for stem cell discoveries. Stem cells separation based on Raman microscopy is possible. Human embryonic stem cells from HES2 cell line can be distinguished as hESCs, hESC-CMs cells, which are human embryonic stem cells and their cardiac derivatives based on glycogen and myofibril as biomarkers. This approach opens the alternative way for their sorting and separation.

RS has found an application in body fluids analysis, which is used in both diagnostics and toxicology studies. In vivo human blood RS can be applied in quantitative glucose sensing with the precision and accuracy comparable with the finger stick devices. Quantitative analysis of fibrinogen in human blood plasma was also shown. As an example of toxicological application, the degradation effect of alcohol on human red blood cells research was carried out.

Toxic Dose

The margin of safety for xylitol is quite high in species other than dogs. In humans, the principal adverse effect is diarrhea caused by xylitol’s slow and incomplete absorption; even this effect requires relatively large amounts to be consumed (>130 g/day).6 In mice the oral median lethal dose (LD50) is at least 20 g/kg.7 In dogs, both published and unpublished data suggest that the ingested doses of more than 0.1 mg/kg may lead to hypoglycemia and that doses of more than 0.5 mg/kg are associated with hepatotoxicity.4 The severity of hepatotoxicity might be idiosyncratic rather than dose dependent because it has been recognized by veterinarians that there are many dogs ingesting xylitol at doses exceeding 0.5 mg/kg that do not exhibit hepatic disease.8 Xylitol toxicity in cats has not been reported in the literature (one anecdotal report only) or thoroughly investigated to my knowledge. Similarly, nearly all that is known about xylitol toxicity in dogs is based on acute exposures and little is known regarding the toxic potential of more chronic exposure to lower doses. A few chronic xylitol exposure studies have been performed in laboratory animals in the former Soviet Union.9 It is interesting to note that these studies often observed little or no effect of feeding diets with considerable xylitol content to canine subjects, suggesting that genetic background or other factors may affect susceptibility.

Calculating an exposure dose can be challenging at times. When the exposure is gum, one must look at the label and determine if xylitol is the first sugar alcohol listed. If the answer is yes, one should base the dose on the total amount of sugar alcohols per piece of gum. If the answer is no, one should assume that there is 0.3 g of xylitol per piece of gum. When the exposure is not gum (like baked goods or powdered xylitol), one should assume that 1 cup of xylitol weighs approximately 190 g.4

ADULT DOGS

Chronic Toxicity

Selamectin has a wide margin of safety in both dogs and cats when used according to the label directions, although, there have been rare reports of neurologic dysfunction including seizures. It has an NOAEL of 40 mg kg−1 day−1 in the dog after 3 months of oral dosing. Numerous studies have shown the safety of selamectin in both young and adult animals. Seven monthly treatments of 6-week-old puppies and kittens at 10 times the recommended dose yielded no adverse effects. Studies have also been carried out in heartworm-positive and breeding animals with no adverse reactions. Selamectin has also been shown to be safe in avermectin sensitive collies when administered at a dosage of 30 mg kg−1 for three monthly doses.

Little is known regarding the health effects in humans chronically exposed to selamectin. However, experimental data suggest little potential for toxicity.

Adverse reactions and side effects

Chloramphenicol has a narrow margin of safety. High doses can produce toxicity in dogs and cats. Gastrointestinal disturbances are rather common. A decrease in protein synthesis in the bone marrow may be associated with prolonged treatment. The effect on the bone marrow is most prominent in cats, especially after 14 days of treatment, but can occur in any animal when exposure is high. Bone marrow suppression in animals is reversible. Idiosyncratic aplastic anemia has been described in humans. The incidence of aplastic anemia is rare but the consequences are severe because it is irreversible. The risk of human exposure led to the ban of chloramphenicol use in food animals. Another problem recognized in dogs is peripheral neuropathy. This causes ataxia, and weakness, particularly in the hind limbs of dogs. Large breed dogs may be more susceptible to this problem. The peripheral neuropathy is reversible if the drug is discontinued.