Besides considering the tumor origin in skeletal involvement with cancer, there are other clinical considerations that may be helpful in estimating prognosis. Clinical prediction of survival (CPS) refers to the clinician’s best prediction of survival based on informal and subjective information. Unfortunately, physicians’ are notoriously optimistic in their estimation of patient survival [67–70], which may explain a reluctance to refer patients to hospice at an earlier point in their illness trajectory.

Performance status intuitively makes sense as a predictor, given that a decline in function occurs as a result of progressive bone involvement. In oncology, both the Karnofsky Performance Status (KPS) and the Eastern Cooperative Oncology Group (ECOG) Performance Status are extensively used to assess eligibility for enrollment in clinical trial or aggressive therapy. The KPS also has demonstrated potential in predicting prognosis [67, 69]. For example, a majority of cancer patients with a KPS score of ≥50 % (i.e., requires considerable assistance and frequent medical care) live more than a month, while the majority of patients who score 10–20 % (very sick, hospitalization necessary, active supportive treatment necessary or Moribund) die within 18 days [71].

The Palliative Performance Scale (PPS) was subsequently developed as a modification of the Karnofsky Performance Status, with the goal of assessing the functional status and survival of patients appropriate for palliative care at end of life [72]. More complex than the KPS, the PPS ranks performance based on ambulation, activity/evidence of disease, ability to care for self, oral intake, and level of consciousness. As with KPS, PPS scores have been shown to correlate with survival, but have been validated primarily in patients who are already in the palliative care setting [62, 83].

In an effort to create a scoring system that included both objective and subjective measures, the palliative prognostic score (PaP) was developed and externally validated in several trials with advanced cancer patients [73, 74]. Based on assessment of patients’ symptoms of anorexia and dyspnea, the KPS, total WBC, presence of lymphopenia, and the clinician’s prediction of survival, mathematical scores are generated and subsequently predict the chances for surviving 1 month. Limitations with the PaP include patients who may have survivals longer than this, and the inclusion of the clinician’s prediction of survival. The PaP also requires a blood sample for determination of the WBC and lymphocyte count, which may not always be desirable or practical at end of life.

One model that relies on a scoring system based on less subjective measures is the palliative prognostic index (PPI). Based on the patient’s palliative performance score (PPS), oral intake, presence of edema, dyspnea at rest, and delirium, patients are divided into one of three group, with survival subsequently estimated in terms of less than 3 or 6 weeks [75]. The PPI has been externally validated; and although the exclusion of the clinical prediction of survival improves the accuracy, its utility is limited to patients with a survival of only a few weeks.

Another model that may have significant utility when predicting prognosis in patients with bone metastases is the number of risk factors (NRF) model . Unlike the other externally validated models discussed, the NRF model has several criteria that are unique to the orthopedic oncology patient population: (1) it was created based on patients referred for radiation, a commonly used palliative treatment option; (2) patients are characterized by the need for radiation to bone versus non-bone sites, and (3) they are further grouped based on breast versus non-breast cancer. The model is quite simple to use, with patients stratified into three prognostic categories based on primary cancer site, presence of bone metastases, and KPS of >60 vs. <60 [76]. Scoring leads to survival predictions of 60 weeks, 26 weeks, or 9 weeks based on the presence or absence of risks.

A variety of normograms have been developed to aid in prognostication, although only the Spain normogram has been externally validated [34]. Unfortunately, its use in the USA is limited as it requires LDH value to be reported in U/L, which is not the typical reporting unit. Although not externally validated, an additional survival normogram based on the PPS, patient age, gender, and tumor origin has been published, is easy to use, and provides a range of best-case/worst-case predictions [77, 78] (Fig. 16.1).

Knowing which prognostic indicator to use when making decisions regarding appropriate therapy for patients with advanced cancer is not clear. For the orthopedic patient, functional status is intuitively predictive. While both the KPS and PPS provide prognostic information, both tests may be more accurate when combined with other measures, such as CPS or laboratory testing [75, 76, 79]. Assessing patients at more than one point in time, noting the rate of decline may also add to accuracy when using tools such as the PPI [78, 80] or the PPS [81, 82]. One multicenter observational study prospectively evaluated the Palliative Prognostic Score (PaP), the D-PaP Score (a modification of the PaP that included delirium as a measurement), the Palliative Performance Scale (PPS), and the Palliative Prognostic Index (PPI). All four models were found to be statistically significant predictive capacity, with the PaP and D-PaP scores being most accurate [38]. Of note, both the PaP and the PPI are predictive for very short survivals; other tools such as the PPS and the NRF model will be more accurate for patients with longer survivals. Finally, in cases where prognosis remains unclear, consultation with both the patient’s oncologist and a primary care physician will be helpful in determining potential surgical interventions in the setting of metastatic cancer.