Giving an accurate prognosis is important for many diseases and is a part of everyday veterinary practice. It is especially important with a diagnosis of cancer, not least because many of the treatment modalities are expensive, time consuming and have the potential to cause side effects. Equally, clients often have an experience of cancer treatment and outcomes for themselves or friends and relatives, and they assume that the prognosis for one type of cancer in the human population mirrors what happens to their companion animal. There are many factors that influence an individual animal’s prognosis and it is important that veterinary surgeons appreciate the usefulness of these factors to help guide the client’s expectations. This article describes how to interpret information obtained regarding the tumours’ histopathological type, grade, stage, and likely response to an appropriate treatment, to provide a prognosis that is as accurate as possible to the client.
Statistics from Altmetric.com
Sue Murphy qualified from the University of Edinburgh in 1986 and initially worked in a general mixed practice for 10 years. She joined the Animal Health Trust in 1997 as a resident, undertaking an MSc in clinical oncology while progressing to become head of oncology and then head of the Centre for Small Animal Studies. She returned to Edinburgh as director of clinics in 2018. She is a European- and RCVS-recognised Specialist in oncology.
Prognosis is defined as the likely course of a medical condition. Every day in practice veterinary surgeons will have the opportunity to discuss with their clients treatment options and prognosis for many and varied clinical conditions. Prognoses can be as diverse as the diseases treated. Diseases such as dilated cardiomyopathy, diabetes mellitus, atopy and benign prostatic hyperplasia have very varied outcomes if untreated, ranging from death to a decreased quality of life. Treatment can have a variable impact on those outcomes. It is exactly the same for animals with cancer.
For all diseases, accurate prognostication should be based on knowledge of the inherent behaviour of the condition and how this is affected by treatment. The effects on prognosis of comorbidities that might affect the ability to deliver the proposed treatment plan, client wishes, cost, the facilities available to the veterinary practice, the proximity to other services such as radiotherapy facilities and the client and patient’s ability to travel have been discounted for the purpose of this article.
So, what influences prognosis when it comes to cancer? Important factors include the tumour’s histopathological type, the stage and grade of the tumour and its likely response to treatment. Some of this information comes from the pathologist or clinical pathologist, some from signalment and a good clinical examination at the time of diagnosis, and some from diagnostic imaging. The prognosis for an individual case comes from interpreting these findings against the literature and understanding the limitations of the literature.
Information from the pathologist
Type of tumour
Cancer is not a single disease – there are over 200 different types. Identification of the tissue of origin will give a broad idea of the tumour’s likely behaviour. This can be achieved either by fine-needle aspiration and cytology or by evaluating an incisional or excisional biopsy. The most important considerations are first, whether the tumour type has the potential to spread, and secondly, how locally invasive the primary tumour is likely to be. A non-invasive tumour that has a very low risk of spread and can be removed easily with adequate surgical margins has the potential to be cured. This type of cancer would include benign tumours such as lipomas, adenomas and haemangiomas, benign mammary tumours, low-malignancy tumours such as basal cell carcinomas, low-grade mast cell tumours (MCTs) and small low-grade soft tissue sarcomas (STSs). At the other end of the spectrum, a tumour that is known to be poorly sensitive to cytotoxic drugs with a high risk of metastasis would have a guarded prognosis even if the primary tumour was resectable. A tumour such as a canine inflammatory mammary carcinoma, which is locally invasive and therefore not amenable to surgical intervention, highly metastatic and chemoresistant, is an example of a cancer that has a very guarded prognosis.
Although the histological type of a tumour may accurately identify its behaviour and therefore prognosis, some common types have a wide range of behaviour. For these tumours, grading can help with prognostication. The National Cancer Institute defines grade as: ‘the description of a tumour based on how abnormal the tumour cells and the tumour tissue look under a microscope. It is an indicator of how quickly a tumour is likely to grow and spread’, and thus informs prognosis.
Tumours cannot be assessed for likely aggressiveness on cytology alone, although it can help. Histopathology is usually needed to show the relationship of cells to normal tissue and the amount of inflammation or necrosis within the tissue, as well as cytological features such as the degree of differentiation of individual cells and the number of mitoses seen within a high-power microscopic field.
There are data that support grading systems for many tumours, including feline mammary carcinomas, canine cutaneous MCTs, canine STSs, canine osteosarcomas, chondrosarcomas, canine squamous cell carcinomas and mammary adenocarcinomas. Grading systems are used particularly to aid decision making for canine MCTs and STSs where there are significant variations in aggressiveness.
Canine cutaneous MCTs
The majority of cutaneous MCTs in dogs are almost benign as they have a low risk of metastasis and can be cured with surgery when adequate margins are taken. However, a significant minority can behave extremely aggressively and spread to the liver and spleen via the draining lymph node and from there elsewhere in the body such that the median survival time (MST) (Box 1) is a matter of months unless there is appropriate treatment. These patients need to undergo both adequate surgery and systemic treatment (chemotherapy or tyrosine kinase inhibitors) to have a chance of survival. As canine cutaneous MCTs are a common tumour that is regularly seen in practice, there is a real need for help in prognostication.
Understanding the jargon
Median survival time
The median survival time (MST) is obtained by lining up the days from (typically) diagnosis to death for each animal in a study in ascending order and taking the middle value. For example, if you have 11 dogs it will be the survival time of the sixth dog in the study. The MST helps to minimise the influence of outliers, that is, dogs that die very early or are long-time survivors, which would happen if the mean was used. Thus, the median is considered a reasonable expectation of survival, but it is worth remembering that 50 per cent of patients survived less than this time and 50 per cent survived for longer. It also can be useful to look at the range, as clients often want to know the best survival time that they can hope for.
Disease-free interval/time to progression
Disease-free interval (DFI) and time to progression (TTP) are terms that are used either when the cancer regrows such that it is clinically evident from a situation where no disease was evident (DFI), or when stable measurable disease enlarges or more lesions are seen (TTP).
Response rate tells us the proportion of animals that respond to a particular treatment. It is usually used when discussing medical or radiotherapy interventions. Response can be defined as a complete, partial or stable disease, and needs to be defined.
Sensitivity is the percentage of cases that are correctly identified as having the described outcome; a high sensitivity means that the number of cases missed by the test is low.
Specificity is the extent to which positives really do have that outcome and not something else.
Until recently, the most useful and widely used grading system was that described by Patnaik and others (1984) over 30 years ago. Under this system, low-grade (grade I) tumours have a good prognosis when treated with adequate surgery alone and high-grade (grade III) tumours have a poor prognosis when treated similarly, with tumour recurrence and metastasis such that the MST is about seven months. Intermediate-grade (grade II) MCTs have, mostly, a good prognosis associated with appropriate surgery alone, but between 5 and 15 per cent metastasise. As the majority of MCTs are grade II tumours, deciding the best way to treat dogs with this grade of tumour is difficult.
A less subjective two-grade system was proposed by Kiupel and others (2011) to try to combat the problems associated with the Patnaik system. In a study using the Kiupel system, the MST was less than four months for high-grade MCTs but more than two years for low-grade MCTs. Repeatability between pathologists was markedly improved (Kiupel and others 2011). Currently, pathologists often assign both a Patnaik grade and a Kiupel grade to MCTs.
STSs can also be divided into low-, intermediate- and high-grade tumours; they are infiltrative but can metastasise. The majority of these tumours are low grade and most of these will be cured by adequate surgery as the metastatic risk is low. In general practice it is common to encounter low-grade, small, discrete tumours that are not fixed to underlying tissue and which have a good prognosis. However, as a common tumour type, there is a good chance a higher grade STS will be encountered. Intermediate-grade tumours have a higher risk of local recurrence and high-grade tumours have a higher risk of both recurrence and metastasis in comparison to low-grade tumours.
Limitations of grading
Although grade is a useful prognostic indicator for feline mammary tubular carcinomas and canine appendicular osteosarcomas, in reality the vast majority of these tumours are high grade so have a guarded prognosis and, in this situation, other indicators such as stage become more important. Grading can also only reliably be performed on a reasonably sized biopsy. Small samples tend to underestimate the grade. Most importantly, many of the grading systems rely on criteria that are subjective, leading to significant variations in grade assigned by pathologists to the same tumour. Recently, there has been a move towards more objective measurable markers of prognosis, which are often related to cell proliferation.
Cell proliferation markers of prognosis
The mitotic index (MI) is defined as the number of mitotic figures identified per 10 high-power fields. It is usually measured on slides stained with haematoxylin and eosin (Fig 1) and given as part of the normal pathology report. Very broadly speaking, the higher the MI, the more concern regarding prognosis. However, the cut-off point between a poor prognosis and a better one identified in studies varies widely between tumour types and sometimes between studies. There is also some variation with pathologists either randomly taking 10 fields or scanning the lesion under a lower power to look for those areas that have most cells undergoing division and counting them, which may account for some of the differences between studies.
Table 1 shows the link between MI and prognosis for some common tumours.
The number of mitoses is a straightforward count but can still be subjective – ‘is that a mitotic figure or not?’ The use of immunohistochemistry to stain for a protein only expressed in cells in all phases of the cell cycle but not in resting cells gives more objectivity in theory. Ki-67 is such a protein (Fig 2).
Although the grade of a canine cutaneous MCT is its most useful prognostic indicator, Ki-67 counts have been used to refine the prognosis, particularly for MCTs where there is a grey area regarding the usefulness of the grade. Such grey areas would be all Patnaik grade II MCTs and Kiupel low-grade tumours where the MI is above 5 (in some studies 5 was used as a cut off for more aggressive tumours).
Low Ki-67 counts seem reliable in identifying Patnaik grade II tumours associated with a good prognosis. However, a high Ki-67 count is not universally associated with a poor prognosis.
Other markers of cell proliferation have been explored for MCTs, including proliferating cell nuclear antigen, agyrophilic nuclear organising regions (AgNORs) and minichromosome maintenance complex component 7 (Fig 3).
Recently, there has been interest in comparing one marker to another to see which is best at predicting outcome; some groups have looked at combining indices to see whether this gives better accuracy. AgNORs are thought to indicate the rate of proliferation so combining this count with a marker of the number of cells in the cell cycle, such as Ki-67, seems to be a rational approach. In one study, a low AgNOR x Ki-67 count was associated with a low chance of recurrence, regardless of surgical margins (Smith and others 2017).
Other prognostic markers identified by immunohistochemistry
Generally, high-grade T cell lymphomas have a worse prognosis than B cell ones and can be identified by immunophenotyping. For some oncologists this information will also inform treatment protocols. KIT patterns have also been used for MCT prognostication.
Other prognostic factors have been assessed in single studies and are not widely used. These are often proteins associated with cell survival, proliferation, angiogenesis, invasion and metastasis, and have been evaluated for several histological types.
Stage is an assessment of the extent of disease – the tumour burden at the moment of examination. The World Health Organization uses the TNM system, in which T refers to the primary tumour, N to the presence or absence of nodal metastatic disease and M describes the presence or otherwise of distant metastasis. Knowledge of the stage not only allows a more informed prognosis but helps with the planning of treatment and provides a baseline to allow the response to treatment to be assessed. It is useful to remember that if a tumour has already spread, the grade of the primary tumour or the presence or absence of other predictive indicators is largely irrelevant.
How do we stage?
Staging a tumour involves assessing the size and invasiveness of the primary tumour and looking for any evidence of the tumour elsewhere. Different tumour types spread in different ways to different organs. Information as to where a particular tumour type metastasises to and the probability of the tumour metastasising to those sites is readily available and should be consulted before performing staging. Staging can be undertaken via a good clinical examination but often involves diagnostic imaging. Sometimes abnormalities need to be explored by cytology.
An example of how stage informs prognosis is given in Table 2.
Clinical signs associated with a higher stage and thus a poorer prognosis are things like the size of the primary tumour, an enlarged, fixed draining lymph node, hepatic or splenic enlargement (with lymphomas and MCTs) and cutaneous metastases (eg, with melanomas and osteosarcomas). It is worth remembering that lymph nodes can be infiltrated with disease and have a normal size or, conversely, can be enlarged because they are reactive. Therefore, it is worth performing a fine-needle aspiration of the node if there is any cause for suspicion or the tumour type has a high rate of spread. It should be noted that the lymph node draining the lesion may not be the one expected. It is well documented that lesions on one side of the body (such as mammary lesions and oral masses) can result in pathology in lymph nodes on the contralateral side or even in different lymph node groups.
The size and appearance of the primary tumour can give useful information. Generally, large, ulcerated, inflamed, rapidly growing, fixed masses are associated with a poor prognosis.
The site of the tumour also has prognostic significance. For common oral tumours, the further rostral a lesion is the better. Tonsillar tumours are associated with a more guarded prognosis, regardless of histological subtype. Cutaneous and subcutaneous MCTs have a much better prognosis than visceral ones. The outcome for canine melanomas is very site dependant. Cutaneous melanomas on the whole have a good prognosis, with the exception of those affecting the digit. Oral melanomas have a high risk of spread, with the exception of well- differentiated, deeply pigmented tumours of the lips. For nasal tumours, epistaxis and facial deformity are associated with a more guarded prognosis.
Limitations of staging
The imaging modality may limit the ability to determine whether a tumour has spread. Three thoracic radiographic views are needed to maximise the chance of seeing pulmonary metastases and computed tomography can identify pulmonary nodules a magnitude smaller than those identifiable by radiographs. It is important to remember that staging using ultrasonography relies on neoplastic tissue having a different echotexture to normal tissue. This is not the case with insulinoma metastases of the liver, which are frequently isoechoic with normal tissue so can be missed. Equally, there are many non-neoplastic lesions that can look suspicious ultrasonographically. All abnormalities should be assessed by fine-needle aspiration and even then it should be appreciated that studies have shown that cytology is not always an accurate way to identify metastatic disease; the gold standard is still histology.
It is difficult to assess studies linking stage to survival unless there is a clear understanding of the way the tumours were staged. It has been shown with canine multicentric lymphomas that cases can migrate from a lower to higher stage depending on whether a clinical examination, routine haematology, biochemistry panel, radiographs, ultrasonography and bone marrow aspiration are performed. In a study by Flory and others (2007), each dog underwent all the staging procedures. If just a clinical examination and routine haematology were used to assess stage, 61 per cent of the animals were assigned to stages I to III, which are traditionally associated with a better prognosis. If the same animals were completely staged using all techniques, only 10 per cent were still considered stage I to III.
It is not uncommon for vets to take blood samples ‘to see whether the cancer has spread’. But how useful is routine haematology or biochemistry? The broad answer is that bloodwork is not very helpful at identifying spread but other prognostic information can be gained from a blood sample; this is described in Box 2.
What information can you get from a routine blood sample?
For both canine and feline lymphomas, finding circulating neoplastic cells identifies advanced-stage disease; however, this type of tumour can be pretty advanced before it is measurable by conventional laboratory methods.
Circulating mast cells in cats with mast cell neoplasia reliably identifies spread but isn’t helpful in dogs.
Anaemia associated with lymphomas, histiocytic sarcomas or multiple myelomas has been identified as a negative prognostic indicator in some reports.
Canine lymphoma-associated hypercalcaemia (which is frequently associated with mediastinal high-grade T cell lymphomas in dogs) is also a negative prognostic indicator. Interestingly, hypercalcaemia in cats with lymphoma is rare.
Elevated serum alkaline phosphatase is a negative prognostic indicator in canine appendicular osteosarcomas, as is hypercholesterolaemia.
The prognosis for many cancers is heavily influenced by appropriate treatment. Untreated, a high-grade B cell canine multicentric lymphoma has a survival time of a matter of weeks, whereas the majority of dogs treated with CHOP (protocols involving cyclophosphamide, vincristine, doxorubicin and prednisolone) will survive with a very good quality of life for months. Broadly speaking, regardless of site, cats and dogs with lymphoma that have an early complete remission to the first cycle of treatment are likely to survive longer compared to those that don’t. Once out of remission, animals can then be treated with rescue protocols. Thus, the prognosis is heavily influenced by the clinician’s knowledge and owner’s willingness to treat.
This is also relevant for tumours where surgery is the mainstay of treatment. The surgical dose can influence prognosis. A tumour taken off with incomplete margins (with the exception of some canine and feline MCTs) will have a more guarded prognosis than one removed with wide, clean margins. A good example is feline injection site sarcomas, where studies have shown that a planned, radical first excision results in a longer survival time compared with marginal excision. In one study, the latest surgical recommendation of 5 cm lateral margins and two fascial planes deep gave a median survival of 901 days and only 14 per cent recurrence. In comparison, more conservative surgery gave a median time to first recurrence (so more than half recurred) of just 79 days (Phelps and others 2011).
After making a diagnosis of cancer, it is important to provide the client with as accurate a prognosis as possible, especially since many of the treatment modalities are expensive, time consuming and have the potential to cause side effects. Although most information gathered for prognosis is useful, there are several things to remember. First, some prognostic indicators are more important than others and have more robust evidence associated with them. Arguably, the most useful information is histological type and stage for solid tumours. For ‘liquid’ tumours (leukaemia and lymphoma), the subtype, grade and then response to treatment with chemotherapy give the most useful information.
Although useful in terms of discussing the course of disease in a population of animals, most prognostic indicators do not provide a crystal ball for each individual case – they provide guidance at best and need interpreting in the light of other prognostic indicators and concurrent disease. Every oncologist knows cases that have either a considerably worse or better outcome than would be expected, and the element of uncertainty associated with any amount of information also needs to be clearly explained to an owner before embarking on a definitive treatment, especially if that ‘treatment’ involves euthanasia.
In Practice has partnered with BMJ OnExamination to host the self-assessment quizzes provided with each clinical article. These can be completed online and found at the end of the online version of each article at http://inpractice.bmj.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.