Prognostic Communication in the Era of Targeted Therapy and Immunotherapy

Abstract

Effectively communicating prognosis to patients with cancer is a key communication task for physicians. It has always been a difficult task and is now becoming more so. Rapid progress in treatment of advanced cancers is transforming a previously terminal illness with its proverbial <6 months of life expectancy into a chronic illness with years of meaningful quality of life ahead. Despite this evolution, the importance of communicating prognosis to enable shared decision-making cannot change. Communication skills for this specific task should be strengthened and refined with practice and toolkits to enable physicians to rise to the unique challenge of discussing prognosis in this rapidly shifting milieu of cancer care. This chapter will first discuss how targeted and immunotherapy have changed the landscape of cancer therapy and complicated prognostication through representative case examples. Secondly, we will outline communication preferences, barriers to prognostication, and tools useful in cancer prognostication. Finally, we will identify techniques palliative physicians and oncologists utilize to convey prognostic information vital to patient decision-making.

1. Introduction

For a patient with cancer, prognosis, the prediction, or estimation of the likely course, has profound effects: preparing to face mortality, making decisions regarding treatments, and hoping for ideal outcomes for the future. Communicating prognosis is central to shared decision-making and autonomy [1, 2], and understanding prognostic information can influence patients’ decisions on whether to receive life extending care [3, 4]. Poor communication therefore can lead to poor outcomes.

Prognostic disclosure in oncology care has been shown to be an essential component of good clinical practice for children [5], adolescent and young adults [6] and on up to the very elderly [7]. Despite the acknowledged importance of communicating prognosis in cancer care, this task has remained a challenge with multiple barriers: physician discomfort in discussing prognosis, patient diversity and other barriers to receiving prognostic information, and patient’s family raising barriers of their own in their belief and acceptance of prognosis. Oncologist’s skills and experiences and comfort level in managing patients’ reactions to prognostic information has emerged as a principal barrier to high-quality prognostic communication [8]. Adding to all these factors are the significant challenges brought about by the rapid advances bringing an array of new and effective treatment for many advanced cancers.

This review will briefly discuss communicating prognosis in the era of progress in cancer treatment ushered in by the identification of molecular targets leading to effective targeted therapy, and the control of advanced cancers by checkpoint inhibitors and immunotherapy. Brief discussion of patient examples to highlight the current profound complexity of discussing prognosis will help appreciate the impact of changes brought about by newer therapy on this communication task. We will provide an overview of prognostic communication preferences and barriers. We will conclude with a discussion of helpful tools and techniques to enable effective prognostic conversation.

2. Evolving landscape of cancer therapy

2.1 Effect of the progress in cancer therapy on communicating prognosis

Oncologists in training 40 years ago very easily saw the natural history of most adult cancers during the 2 or 3 years of their fellowship program. There were very few effective treatments for the most common advanced stage cancers. In that period, discussing prognosis with patients who had advanced cancer caused emotional distress in oncologists and their patients and this was a significant barrier to effective conversation about prognosis. Contrast that with what the current trainee faces in this era of targeted and immunotherapy: 50 or more new drugs for cancer treatment have been approved in both 2020 and 2021 [9, 10] and a significant decrease is now evident in cancer mortality [11] contributing to an increased lifespan [12]. Therefore, advanced cancers now are not always rapidly fatal and decision-making regarding treatment forces even the expert hematology and medical oncology physicians to pause before rendering prognostic opinions. The pause is not only to assess the impact of performance status and co-morbidities on prognosis, but also to: 1) ponder and choose from a bewildering array of treatment choices; 2) to consider the likelihood of rapid response and degree of complete remissions possible in the “exceptional responders;” 3) to consider the quality of life of patients who may have to learn to live with their cancer on maintenance therapy; and 4) to consider if the patient they face now will one day be monitored for treatment-free remissions. All these points deeply influence how we should think about discussing prognosis in patients with advanced cancer.

2.2 Prognostic communication in patients with solid tumors—Case-based discussion

2.2.1 Case #1: ALK mutated lung cancer

Take for example, a 53-year-old female with diabetes who reported a dry non-productive cough starting in March of 2017. She was evaluated by her primary care physician and found to have a mass of her right lung. Inadequate health insurance resulted in delays and a biopsy was obtained only in August 2017 confirming poorly differentiated high-grade neuroendocrine carcinoma with metastatic disease to the lumbar spine. She received palliative radiation to the spine, then started on standard systemic chemotherapy with cisplatin and etoposide but progressed after three cycles. She was then switched to standard second line immunotherapy in November. Restaging scans after three cycles revealed mixed response. At this point she was following the typical natural course of lung cancer, and without new effective therapy her prognosis would have been grim. However, molecular profiling of her tumor demonstrated an ALK mutation. She was transitioned to the targeted agent ceritinib, and experienced a near complete response (Figure 1). She has remained on therapy without adverse events and no evidence of disease to date far exceeding the prognostic predictions at the time of second line therapy.

Figure 1.

A. Representative axial and coronal cross sectionals pretreatment with Alk inhibitor. B. Representative axial and coronal cross sectionals posttreatment with Alk inhibitor demonstrating near complete response.

2.2.2 Case #2: Metastatic non-small cell lung cancer

Consider a 64-year-old female with adenocarcinoma of the lung with metastatic disease to the retroperitoneum and pancreatic tail who was treated on a clinical trial with induction chemotherapy of carboplatin, paclitaxel, bevacizumab and atezolizumab resulting in rapid and dramatic response; followed by Atezolizumab maintenance with continued control of disease when she developed an encephalopathy syndrome due to the atezolizumab and treatment was discontinued. Off therapy, she developed rapid onset of progressive metastatic disease to bone causing large destructive lesions of her left hemipelvis requiring orthopedic surgery to stabilize her hip. She then completed radiation therapy to the involved area and then began maintenance doses of mono-immunotherapy with nivolumab without recurrence of encephalopathy. She has remained on maintenance immunotherapy for greater than 5 years with no evidence of disease enjoying an excellent quality of life with near normal performance status.

2.3 Prognostic communication in patients with hematologic malignancies—Case-based discussion

Prognostic information has generally been better studied in patients with solid tumors. Patients with hematological malignancies faced uncertain illness trajectory, treatment choices associated with risk of severe toxicity, but also a persistent chance for cure from stem cell transplant or CAR T cell therapy for the select few eligible for such options [13, 14]. Targeted therapy and immunotherapy have dramatically improved outcomes for patients with hematological malignancies [15].

2.3.1 Case #1: Multiple myeloma

A healthy man presented at age 68 with new-onset mid back pain. Imaging showed a deformity at T8 arising from a lytic lesion, along with lesions in other bones. Bone marrow biopsy demonstrated 40% clonal plasma cells, with FISH showing translocation (11, 14). Treatment was initiated with bortezomib, lenalidomide, and dexamethasone. After 2 cycles he developed grade 2 peripheral neuropathy, and bortezomib was discontinued. After 5 additional cycles of lenalidomide and dexamethasone, the monoclonal protein dropped more than 90% relative to pre-treatment baseline, meeting criteria for a very good partial response. He then received an autologous stem cell transplant, which he tolerated without major complications. Post-transplant, the monoclonal protein was undetectable, and lenalidomide maintenance was initiated. After just over two years of maintenance therapy he began to complain of worsening back pain. Imaging showed enlargement of several bony lesions with a new compression deformity at L1. His performance status remained excellent.

This example illustrates a common clinical situation of symptomatic relapse of multiple myeloma after stem cell transplant. Prior to 2012, the options for such a patient would have been limited. In 2022, the array of options in the same clinical situation is broad. Pomalidomide, carfilzomib, and daratumumab are all widely used drugs offering high response rates, alone and in numerous combinations, and all have mild toxicity profiles. These drugs can be combined to make several lines of therapy, even in patients with less-than-optimal organ function and performance status. Even after a patient becomes refractory to these drugs, newer salvage options are available, including selinexor, belantamab mafodotin, and elotuzumab. Several investigational drugs are in advanced stages of development and will likely add more treatment options in the near future. CAR-T therapy is commercially available as of 2021, and offers very high response rates, while incurring high cost and complexity of treatment. Finally, the patient’s 11;14 translocation makes him a candidate to receive venetoclax, which has shown activity in this subgroup. With access to these treatments, survival for five years or longer after a first relapse is becoming routine. However, the likelihood of achieving response and the duration of response tend to drop with each successive line of therapy. It is rare to exhaust all available lines of therapy, but the tradeoffs between burdens and benefits of treatment shift progressively though the course of disease.

2.3.2 Case #2: Acute myeloid leukemia

A 73-year-old woman with diabetes, hypertension, stage IV chronic kidney disease, and mild dementia presented with nausea, vomiting, and chest pain. Laboratory tests on admission showed leukocytosis of 22,000/μL along with moderate anemia and thrombocytopenia. Bone marrow biopsy was diagnostic for acute myeloid leukemia (AML).

Management of AML in the elderly is a longstanding challenge as the benefits of intensive chemotherapy are outweighed by the risk of infections and other complications. Historically these patients were treated with lower-intensity cytotoxic regimens, which could prolong survival by a few months at best.

Hypomethylating agents (azacitidine and decitabine) in combination with the BCL-2 inhibitor venetoclax have transformed the care of the elderly AML patient giving them a median survival of 14.7 months [16]. Although the combination can be tolerated by frail elders, it is not free from toxicity. The patient described above, despite kidney disease, early dementia, and limited functional status, has no absolute contraindications to such treatment, though these factors may negatively affect tolerability. Deciding whether to start combination therapy, a single agent, or best supportive care is therefore a complex and individualized decision.

2.4 Effect of the progress in cancer therapy on clinical practice

These cases serve to illustrate the high level of complexity that exists from response to later lines of therapy, to associated toxicities, to anticipated quality of life on treatment. Targeted and immunotherapies are associated with both palliative benefits in terms of pain/symptom control and improved survival benefits compared to former cytotoxic chemotherapy options (Table 1). Additionally, these therapeutics are associated with fewer and less severe toxicities such that patients can remain on treatment for extended periods of time.

Disease	Treatment options in 2010	Prognosis circa 2010	Major developments	Prognosis in 2022
Metastatic NSCLC	Doublet or triplet cytotoxic chemotherapy	mOS approximately 10–12 months [17, 18], 5-year survival 8% [19]	EGFR agents	Targeted Therapy mOS (mo) [20]: EGFR 18–38 [21, 22]
			ALK agents	ALK 47-not reached
			ROS1	ROS1 24-not reached
			BRAF	BRAF 17
			RET	KRAS 12.5
			MET	RET, MET,NTRK data immature
			NTRK	Targeted Therapy: 5-year survival 83% [23]
			KRAS immunotherapy	Immunotherapy: Median OS 22 months (treatment naïve), 5 year survival 23% [24]
AML (older and less fit patients, not fit for intensive chemotherapy)	Low-dose cytotoxic chemotherapy	Median OS <6 months; 1-year survival 28% [25]	Hypomethylating agents, BCL-2 inhibitor	Median OS 15 months; 2-year survival 74% [16]
Multiple myeloma	Lenalidomide or bortezomib, combined with steroids and/or chemotherapy	Overall survival 1–2 years [26]	Multidrug combinations including: 2nd generation immunomodulatory drugs	OS unknown, but likely 5 years or more
			2nd generation proteosome inhibitors
			monoclonal antibodies
			CAR-T cellular therapy

Table 1.

Evolution of prognosis in representative case examples.

These factors in combination with the hope for durable responses have resulted in a shift in referral patterns for patients with advanced and metastatic disease. In a 2019 study, a trend toward fewer hospice referrals and increased subacute rehab referrals from inpatient oncology units was noted with nearly two-thirds of patients never receiving additional cancer therapy [27]. In another retrospective study of deceased patients who had received immunotherapy, two-thirds had received immunotherapy in the last 90 days of life [28]. Notably, patients who had received immunotherapy in the last 30 days of life received less than 3 doses, had a poor performance status, had lower hospice enrollment, and higher rates of dying in the hospital [28]. Although extraordinary and durable responses are seen, these two studies emphasis the prognostic dilemma, as the majority of patients are likely to follow a different natural course. Therefore, assessing patient preferences for information sharing and goals of care is essential.

3. Prognostic communication preferences in oncology

3.1 Prognostic communication clinician factors

Communication is a specialized skill. Indeed a review of the literature of prognosis related communication in advanced cancer patients suggests that it is useful to divide oncologists into three groups with respect to their ability to engage in meaningful high-quality communication with their patients: Highly skilled oncologists need organizational support and can serve as mentors; moderately skilled oncologists may benefit from targeted skills training; and lower skilled oncologists may benefit from pairing with high level communicators or utilization of supportive programs to facilitate effective communication [8]. Discussing prognosis does include asking patients if they wish to receive this information and being prepared to respect their decision and re-engage periodically.

Communication training and skills are common barriers reported by clinicians. Inadequate training can result in brief, vague, or total avoidance of prognostic discussions. Conversely, physicians may use jargoned language with the intent to deliver accurate information or the uncertainty of prognosis, however no additional clarity is provided to the patient. Additionally, these conversations may be viewed as time consuming in the context of a busy oncology practice.

The emotional nature of prognosis contributes to additional clinician discomfort with delivering bad news and managing patient responses. Despite data suggesting otherwise, many physicians feel prognostic conversations decrease patient hope and create a less favorable provider-patient relationship [8, 29]. This poses a significant barrier as many oncologists develop a personal bond with their patients.

Finally, physician experience plays a significant role. Rapid progress in treatment has led to the finding that hematologic oncologists with fewer years of practice are less likely to engage in prognostic discussion [30] compared with younger physicians caring for patients with solid tumors who were significantly more likely to discuss prognosis than their older colleagues [31]. This reflects a higher level of “information uncertainty” and less confidence among junior hematologic oncologists as a major prognostication challenge for hematological malignancies [30].

3.2 Prognostic communication patient preferences

For patients and families, having prognostic information influences treatment preferences, decreases uncertainty and helps them to plan ahead for both personal and healthcare matters [32]. The majority of patients prefer to have prognosis communicated and nearly universally, they want accurate and honest information [33]. While some variation in preferences is influenced by age and sex, underlying cancer type and treatment goals do not impact patient preference [29]. In this era then, communication of prognosis with the elderly person with advanced cancer deserves special mention: Communicating prognosis in the elderly, especially in the very elderly (>85), or frail elderly is important for establishing patient centered goals of care and advance care planning. Fear, grief, and anxiety are common factors contributing to patient related barriers in prognostic communication and therefore invoking a patient’s preference for information sharing remains essential [29]. Additionally, language and education barriers contribute to prognostic misunderstandings [29].

3.3 Impact of care giver preferences on prognostic communication

While important on an individual level, cultural and community factors often guide care giver communication preferences. For instance, South American and Asian cultures are less likely to believe patients should be told about a terminal prognosis [29]. This discrepancy between patient and family preferences can contribute to prognostic communication barriers. Discussing prognosis often involves a family member as the patient may rely increasingly on a family member for decision making and care—and the patient, especially if elderly, is often more receptive than the family when hearing prognostic information.

4. Prognostic tools

4.1 Clinician prediction of survival (CPS)

Clinician prediction of survival (CPS) is a quick and convenient estimate of survival based on clinical experience and intuition. While CPS is convenient, it is inaccurate and subject to biases with consistent over estimation of survival. Literature review suggests CPS is accurate 20–30% of the time with ~80% of prognostic errors being overoptimistic [33]. Overestimation may lead to overtreatment and delayed referrals to palliative care. Conversely, while less common, underestimation can result in undertreatment and premature referral to hospice or supportive services [33]. CPS can be classified by several approaches. The most common technique utilized by clinicians is a temporal approach which attempts to answer “How long will the patient live?” or providing an estimated duration of survival [32]. A probabilistic approach estimates “What is the probability of survival in a specific time frame?” (i.e. 30% chance of being alive in 6 months) [32]. Alternatively, the surprise question is utilized to determine “Would I be surprised if a patient died in the next year?” [34]. The surprise question significantly improves clinician accuracy in predicting poor prognosis. In order to provide the most accurate prognosis several models have been developed to facilitate clinical communication. While these validated models provide more objective prognosis, they do not necessarily capture the nuances of an individual’s health.

4.2 Palliative prognostic score (PaP)

The palliative prognostic score was initially developed in Italy to be used as an estimate of short-term survival [35, 36]. PaP combines CPS, Karofsky Performance Score (KPS) and five clinical/laboratory variables including performance status, dyspnea, anorexia, leukocyte count, lymphocyte percentage (Figure 2) [37]. Each criterion is assigned a score which are then summed to generate a numerical score (0–17.5). The values are then stratified into three groups according to a 30-day survival probability. The PaP has been validated in advanced cancer settings and shown to be accurate irrespective of cancer type [38]. In an interesting validation study, the inclusion of CPS as a criterion enhanced the accuracy [39].

Figure 2.

Palliative prognostic score.

4.3 Palliative performance scale (PPS)

The PPS, originally developed in 1996, is a reliable, validated tool which uses five observer-rated domains correlated to the Karnofsky Performance Scale (100–0%) with scores in 10% decrements (Figure 3) [40, 41, 42]. The domains include ambulation, activity level/evidence of disease, self-care, intake, and level of consciousness. PPS scores are determined based on a “best fit” while reading downward through a single domain and then across the remaining domains left to right. If several domains are categorized at one level and other domains at a higher or lower level, clinical judgement and leftward precedence is used to determine the more accurate score. While the PPS may be used for different purposes, it is a key tool for quickly communicating a patient’s functional level. More importantly, the PPS is a valuable prognostic tool as scores correlate with actual survival and median survival time for cancer patients in the ambulatory setting thereby allowing estimates in terms of days, weeks, months, and years (Figure 4) [43, 44].

Figure 3.

Palliative performance scale version 2 (PPSv2).

Figure 4.

Survival curves of cases with different PPS.

4.4 Palliative prognostic index

The Palliative Prognostic Index(PPI) was originally developed in Japan for hospice in patients with advanced cancer [45]. PPI utilizes PPS, oral intake, and the presence or absence of dyspnea, edema, and delirium (Figure 5) [46]. Criteria are assigned a numeric score and the total is stratified into one of three groups, predicting survival of shorter than three weeks (PPI score greater than 6), shorter than six weeks (PPI score greater than 4), or more than six weeks (PPI score less than or equal to 4) [47].

Figure 5.

Palliative prognostic index.

4.5 Terminal cancer prognostic index

The Terminal Cancer Prognostic score (TCP score) is based on the weighted scores of three predictors (diarrhea, anorexia, and confusion), which were determined in multivariate analysis to be independent predictors of survival for terminally ill cancer patients (Figure 6) [46, 48]. The scores are then used to differentiate into a prognostic group. While developed in a prospective study, this tool has not been adequately validated and is not as commonly utilized [46].

Figure 6.

Terminal cancer prognostic index.

4.6 Poor prognostic indicator

In a similar pattern to the TCP score, the poor prognostic indicator is a score developed based on dysphagia, cognitive failure, and weight loss in the last 6 months [46, 49]. The score has a PPV of 0.76 at estimating 4-week mortality for patients with advanced cancer admitted to a palliative care unit. This prospective study demonstrated that the poor prognostic indicator was as effective in predictions of survival as two skilled physicians, although this tool has also not been adequately validated [49].

4.7 Charlson comorbidity index (CCI)

First developed in 1987 as a weighted approach to assess risk of death within 1-year of hospitalization with specific comorbid conditions [50]. The index assigns value to 19 medical conditions (i.e., diabetes, heart failure, peripheral vascular disease, chronic pulmonary disease, liver disease, renal disease, hemiplegia, etc.) according to presence and severity. The sum is used to estimate the risk of death. CCI is the most common comorbidity assessment utilized and has been validated in multiple solid and hematologic malignancies [51, 52, 53, 54].

4.8 Prognostic tools in clinical practice

As mentioned, CPS is the most used method; however, the accuracy is bolstered significantly when used in combination with other more objective tools. The PaP likely has the greatest potential utility however it is unclear why this tool is not used more universally other than lack of familiarity. PPS and CCI are the most broadly used tools in both oncology and palliative care settings and have the most robust data in terms of validity in patients with cancer. Both tools also assist with longer term prognostication which is increasingly important as cancer care transitions from anticipated acute mortality to a chronic illness with associated morbidity and mortality.

5. Prognostic communication techniques

While developing and communicating prognosis is critical for patient care, evidence-based recommendations for these processes are limited. In 2005, a Working Group of the Research Network of the European Association for Palliative Care found that prognostication is feasible with the use of certain clinical tools [55]. The working group found the strongest evidence of prognostic correlation with the use of clinical prediction of survival in conjunction with performance status, symptoms associated with cancer anorexia-cachexia syndrome (weight loss, anorexia, dysphagia, and xerostomia), dyspnea, delirium, and some biologic factors (leukocytosis, lymphocytopenia, and C-reactive protein) [55]. More recently in an opinion statement, similar factors were identified as key to more accurate prognostication [56]. In addition to using validated prognostic tools to complement clinical judgement, the authors recommended seeking multi-professional prognostic estimates, formulating and recording prognosis estimates in order to better cultivate prognostic skill, and receiving training in advanced communication skills to better deliver prognostic conversations [56].

Every cancer is a complex clinical phenomenon, whose outcome is influenced by many variables: the age, underlying health, and motivation of the patient; their social support system; the availability of advanced treatments, clinicians’ ability to choose and manage these treatments, patient’s ability to access and afford these treatments and tolerate their side effects; and critically, the biology of the cancer itself. Even in an age of molecular and genetic tumor profiling and validated prognostic models, an estimate of prognosis is an educated guess. Clinicians must make this guess in the most accurate and objective way possible, and then use communication techniques that effectively convey the information to the patient and caregivers, while being sensitive to their receptiveness and ability to process this information.

Surveys have consistently shown that majorities of patients prefer to receive accurate and honest prognostic information, [33, 57] including average and best-case survival outcomes [58]. However, this is not universal: there is considerable variation in how much prognostic information patients and caregivers wish to receive [13, 59]. This can change over the course of illness: as the disease becomes more advanced, the patient may want less detailed prognostic information, while caregivers may need more. Therefore, prognostic discussion calls for an individualized approach, which elicits permission from patients and caregivers to discuss prognosis, and uses explicit questions and implicit cues to determine how much prognostic information to impart [3, 60, 61]. Information needs and preferences should be reassessed any time there is a significant change in the condition, prognosis, or living circumstances of the patient.

An effective discussion of prognosis is not limited to a prediction of medical outcomes, but should integrate affective communication including expressions of support and empathy, demonstration of expertise, reassurance of non-abandonment, and reinforcement of a collaborative relationship [60, 62]. Assuming permission has been given to offer predictions regarding outcome, it is important to utilize the opportunity to openly and directly discuss prognosis [61]. Clinicians should be as transparent as possible about the basis for the estimate – multiple studies, a single clinical trial, or simply an estimate based on personal clinical experience – and about the associated uncertainties.

Presenting a range of possible outcomes is a useful technique to convey the scope of uncertainty, and is preferred by patients over more narrow predictions [63]. Effort should be taken to counteract the tendency of both clinicians and patients to focus on the most favorable potential outcome. One way to frame this is surprise: describing a relatively favorable outcome but stating that it would be surprising, implying that it should not be assumed. The surprising outcome can then be contrasted against other outcomes which are expected or feared. This is one form of the best case/worst case framework, which invites hope for a favorable outcome, while encouraging preparation for a more adverse one. This approach was first studied in the context of high-risk surgery, [64, 65] but it is broadly applicable in any situation where plausible outcome scenarios differ widely. In addition to best and worst, a third scenario for the typical or average case, but it is important to emphasize that this is not a prediction, and that the best and worst outcomes remain possible.

There is no one-size-fits-all approach to improving communication skills. For delivering prognostic information, collaboration with other cancer clinicians, oncology nurses, and palliative care specialists is encouraged [66]. REMAP (Reframe, Expect emotions, Map out patient goals, Align with goals, and Propose a plan) has been suggested as a framework for goals of care conversations [67].

Indeed viewing communication interactions from a procedural lens (procedures require specialized skills, have component steps, are intentional in purpose, and have pause points and a known set of complications) may improve quality of prognostic communication [68]. In discussing the use of prognostic tools in communicating prognosis, it is useful to quote from a paper by Lakin et al., entitled, Timeout Before Talking: Communication as a medical procedure: “Skilled communication requires nuance, adjustments, and careful thought, in complex interpersonal interactions. Just as surgery is a technical intervention and a practiced art form, communication procedures require both thoughtful structure and flexible skill. When a communication task is deconstructed, it can be better applied, taught, legislated, and researched, ultimately allowing for iterative improvement of this foundational medical practice” [68].

6. Conclusions

Prognostication is an evolving science. As newer therapeutics extend and improve the expectations of critical illness management, transparency and accuracy are key factors of value to patients. For clinicians, continuing education in communication and routinely rehearsing difficult conversations may promote easier and more transparent conversations. Utilization of prognostic tools in combination with clinician prediction and interdisciplinary consensus improves overall accuracy in uncertain circumstances. Prognostic discussions should be viewed by the clinician and patient as fluid. Routinely re-evaluating and modifying prognostic estimates based on current individual circumstances, changing health status, and therapeutic advances are critical to support patient centered decision making.

Conflict of interest

The authors declare no conflict of interest.

Notes/thanks/other declarations

Figures 2, 5, and 6 reprinted from Stone PC, Lund S. Predicting prognosis in patients with advanced cancer. Annals of Oncology. 2007;18(6):971-976, with permission from Elsevier.

Figure 3 reprinted with permission from ©Victoria Hospice Society, Canada. www.victoriahospice.org.

Figure 4 reprinted from Morita T, Tsunoda J, Inoue S, Chihara S. Validity of the palliative performance scale from a survival perspective. Journal of Pain and Symptom Management. 1999;18(1):2-3, with permission from Elsevier.