Should patients with advanced, incurable cancers ever be sent home with total parenteral nutrition?

A single institution's 20-year experience




Home total parenteral nutrition (TPN) can be lifesaving and life sustaining for some patients. However, in patients with advanced, incurable cancer, its role is controversial. A retrospective study was conducted to explore whether home TPN was associated with long-term survival (≥ 1 year) in patients with metastatic disease and to identify predictive factors to enable its judicious use.


The records of all adult patients with incurable cancer were identified between 1979 and 1999. Records were reviewed in depth for survival from TPN initiation to death and for a variety of demographic and clinical factors.


Fifty-two patients were identified. Their median age was 56 years (range, 18–83 years), and 30 (58%) were women. Malignant diagnoses included carcinoid/islet cell tumor (n = 10), ovarian carcinoma (n = 6), amyloidosis/multiple myeloma (n = 6), colorectal carcinoma (n = 5), sarcoma (n = 5), pancreatic carcinoma (n = 4), gastric carcinoma (n = 3), lymphoma (n = 2), pseudomyxoma peritonei (n = 2), and other (n = 9). TPN was initiated for the following reasons (indications are not mutually exclusive): alimentary tract obstruction (n = 20), short bowel syndrome/malabsorption (n = 16), fistula (n = 11), dysmotility (n = 3), nausea/emesis (n = 2), anorexia (n = 2), and mucositis (n = 1). The median time from initiation of TPN to death was 5 months (range, 1–154 months). Sixteen patients survived ≥ 1 year. TPN-related complications included 18 catheter infections (1 per 2.8 catheter-years), 4 thromboses, 3 pneumothoraces, and 2 episodes of TPN-related liver disease. Tumor grade, the interval between diagnosis of metastatic disease and initiation of TPN, the presence of prominent cancer symptoms, and the administration of cancer therapy after TPN were not associated in any way with overall survival.


The initiation of home TPN can be associated with long-term survival in very select patients with incurable cancer, and complication rates with its use appear acceptable. However, the judicious use of home TPN in this setting requires careful clinical assessment on a patient-by-patient basis. Cancer 2005. © 2005 American Cancer Society.

Total parenteral nutrition (TPN) can be a lifesaving and life-sustaining intervention for some patients.1 In patients with advanced-stage, incurable cancer, however, the role of TPN remains controversial. Multiple, randomized clinical trials comprising patients with advanced-stage cancer have failed to demonstrate any clinical benefits.2–4 The poor outcomes observed in these trials culminated in a consensus statement from the American College of Physicians. This statement advised that the routine use of parenteral nutrition should be discouraged in patients undergoing chemotherapy and that when it is used in cancer patients with malnutrition, physicians should consider the possibility of increased risk.5 This statement and the clinical trials that led to its issuance have curbed the use of TPN in patients with metastatic, incurable disease.

Occasionally, however, patients with advanced-stage, incurable cancer in the United States and in other parts of the world continue to receive TPN, and, at times, are discharged from the hospital with home TPN. Although many of these patients suffer complications from this intervention and live only a few months, some appear to live an extended life, presumably as a result of TPN.6–10 Such anecdotal reports and clinical series prompted this retrospective study.

The current study had three goals. First, we described a single institution's experience with TPN in patients with advanced-stage, incurable cancer in the outpatient setting. The outpatient setting was chosen because long-term survivors are more likely to be identified in this setting. We also excluded the occasional scenario in which a patient with advanced-stage cancer receives transient TPN in the hospital as end-of-life issues are being settled. For this reason, also, we focused on home TPN by asking the following questions: Who received TPN at home? What malignancies did they have? What were the clinical circumstances that prompted its initiation?

Second, we explored survival in the setting of home TPN. Survival after cessation of energy intake can be quite variable, but usually falls within the range of only a few months, even among healthy individuals with a favorable initial nutritional status.11 We determined whether there were any long-term survivors who lived for ≥ 1 year after initiation of TPN.

The third goal of our study was to identify factors that might predict a longer survival among TPN-treated patients and thereby, perhaps, promote its judicious use. Our hypotheses relevant to this third goal were as follows: 1) patients with low-grade tumors or clinically indolent malignancies live longer than patients with more aggressive malignancies; 2) patients with prominent cancer symptoms, such as pain or shortness of breath, at the time of TPN initiation do not live as long; and 3) patients who receive cancer treatment (e.g., chemotherapy, hormonal therapy, or radiotherapy) live longer than those who do not. To address these 3 goals, we examined the Mayo Clinic's 20-year experience with TPN in patients with incurable, metastatic disease in the outpatient setting.


The institutional review board of the Mayo Clinic (Rochester, MN) reviewed and approved all aspects of the study before its initiation. Thereafter, all adult patients with metastatic disease who had TPN initiated at the Mayo Clinic between 1979 and 1999 and who continued to receive TPN in the outpatient setting were identified and included in the study. Because all patients must undergo a consultation with the Home Total Parenteral Nutrition Service at our institution before initiation of outpatient parenteral feeding, the records from this clinical service were utilized to identify patients.

After identifying all potentially eligible patients, inpatient and outpatient medical charts were reviewed. The following data were recorded: age at the time of TPN initiation; gender; tumor type and grade; date of diagnosis of metastases; reason for TPN; whether a surgical versus nonsurgical specialty initiated TPN; date of initiation and cessation of TPN; symptoms present at TPN initiation; whether cancer therapy was given after TPN initiation and, if so, what type; TPN-associated complications; reason for stopping TPN; and date of death. Tumor grade may not have the same impact across tumor types, but its use in this manner seemed reasonable in view of a lack of other methodology for predicting the aggressiveness of such a variety of malignant diseases.

Patients were deemed long-term survivors based on a survival period of ≥ 1 year. The 1-year mark was chosen because earlier studies have been unclear regarding survival rates during starvation or after cessation of energy intake.11 However, a ≥ 1-year survival seemed incontestably favorable and unlikely to be completely attributable to the body's ongoing consumption of energy stores.

Members of the Mayo TPN group (JB, DK), who prescribed TPN to all outpatients, reviewed all the clinical data and corrected any discrepancies between the general outpatient and inpatient medical record and written TPN records maintained by the Home Total Parenteral Nutrition Service.

In general, the initiation of home TPN at our institution involved nocturnal infusions of an approximate duration of 12 hours. Laboratory monitoring occurred weekly and then gradually shifted to longer intervals based on stabilization of laboratory values. The home TPN nursing and physician staff initially followed patients weekly and then gradually less frequently over time.

Demographic data are presented descriptively. Exploratory analyses were undertaken to assess whether any of the following factors predicted overall survival: 1) tumor grade and interval between diagnosis of metastatic disease and initiation of TPN 2 (separate analyses; 2) the presence of prominent cancer symptoms at TPN initiation; and 3) the administration of cancer therapy after TPN initiation. Kaplan–Meier survival curves were constructed. Log-rank tests were used to make comparisons between groups, as defined by the above four factors. For patients who remained alive, the date of last follow-up was recorded, and subsequent data were censored. P < 0.05 was statistically significant.


Fifty-two patients were identified over this 20-year period. As a general reference, 372 patients, inclusive of the 52 mentioned above, received TPN at home during this interval. Among these 52 patients with cancer, the median age at the initiation of TPN was 56 years (range, 18–83 years). There were 30 women (58%) and 22 men (42%). The most common cancer within this group of patients was carcinoid or islet cell tumors (Table 1), and the most common reason for TPN initiation was obstruction of the alimentary tract (Table 2). Thirty-seven patients (71%) had lost ≥ 4.5 kg before initiation of TPN, although the interval during which this weight loss occurred was not always specified in the medical record. A surgical service initiated the TPN in consultation with the Home Total Parenteral Nutrition Service in 25 patients (48%).

Table 1. Malignancies (n = 52)
CharacteristicsNo. of patients
Carcinoid/islet cell tumor10
Ovarian carcinoma6
Amyloidosis/multiple myeloma6
Colorectal carcinoma5
Pancreatic carcinoma4
Gastric carcinoma3
Pseudomyxoma peritonei2
Breast carcinoma1
Jejunal carcinoma1
Vaginal carcinoma (clear cell)1
Cervical carcinoma1
Testicular carcinoma1
Table 2. Circumstances Surrounding the Initiation of TPN
Reasons for TPNNo. of patientsa
  • TPN: total parenteral nutrition.

  • a

    All numbers in each group do not add to 52, as some patients belong to multiple categories.

Alimentary tract obstruction20
Short bowel syndrome/malabsorption16
Weight loss before TPN 
 ≥10 lb37
 <10 lb6
Was it known that the patient had cancer when TPN was started? 
 Unable to tell from chart1
Specialty that initiated TPN 
 Oncology or hematology (medical)14
 Gastroenterology (medical)9

Tumor grade was available for 33 patients, with 13 patients (39%) having low-grade tumors (Grade 1 or 2) as opposed to high-grade tumors (Grade 3 or 4). Thirty-two patients (62%) had an interval of < 1 year between their diagnosis of metastatic disease and the initiation of TPN. Such data were unavailable for 1 patient, and for all others this interval was > 1 year. Thirteen patients had symptoms of pain or shortness of breath at the time of TPN initiation. Fifteen patients received chemotherapy after the initiation of TPN; one received radiotherapy; and two received hormonal therapy.

The median time from the initiation of TPN to death was 5 months (range, 1–154 months). Sixteen patients survived ≥ 1 year. Survival data were censored for 4 patients, 1 at 154 months, another at 2 months, and 2 at 1 month (Table 3 and Fig. 1).

Table 3. Long-Term Survivors (≥1 Year)
Age (yrs)GenderCancerReason for starting TPNReason for stopping TPNCancer therapy after starting TPN?Survival (mos)
  1. F: female; M: male; TPN: total parenteral nutrition.

62FCarcinoidMalabsorption/short bowel syndromeOngoingNo154 (censored)
64FCarcinoidMalabsorption/short bowel syndromeDeathChemotherapy74
18MSarcomaMucositisNo more needChemotherapy36
45FOvarian carcinomaBowel obstruction, short bowel/malabsorptionDeathChemotherapy35
55FCarcinoidBowel obstructionDeathNo12
44MAppendiceal cystadenocarcinomaBowel obstructionDeathChemotherapy22
72FPancreatic carcinoma (adenocarcinoma)DysmotilityNo more needUnknown26
45FCervical carcinomaBowel obstruction, fistulaDeathNo12
72FCarcinoidShort bowel/malabsorptionDeathNo20
20MSarcomaGastric outlet obstructionDeathChemotherapy18
74MCarcinoidBowel obstruction, short bowel syndromeDeathNo77
65FOvarian carcinomaFistulaNo more needChemotherapy19
56MLymphomaMalabsorption, bowel obstructionUnknownUnknown26
43FAmyloidosisMalabsorptionNo more needChemotherapy19
29FEsthesioneuroblastomaShort bowel syndromeNo more needRadiotherapy75
Figure 1.

Within the 52-patient group, the median survival after initiation of total parenteral nutrition was 5 months (range, 1–154 months).

For nearly all other patients, the date of TPN cessation was virtually the same as the date of death, and death was the most common reason for TPN cessation (Table 4). The few notable exceptions consisted of 1) a patient with sarcoma who received TPN because of mucositis and lived for almost 3 years after cessation; 2) a patient with amyloidosis who received TPN because of malabsorption and lived approximately 18 months after cessation; 3) a patient with ovarian carcinoma who started TPN because of an enterocutaneous fistula and lived slightly > 1 year after cessation; 4) a patient with pancreatic carcinoma who begun TPN because of motility problems and lived almost 2 years after its cessation; and 5) an esthesioneuroblastoma patient who had dysmotility problems, stopped and restarted TPN, and overall lived for a few years before having to reinitiate TPN.

Table 4. Reasons for Stopping TPN (n = 52)
ReasonsNo. of patients
  1. TPN: total parenteral nutrition.

No need9
 Fluid overload1

Among the potential predictive variables examined, none was significantly associated with improved survival. To address our first hypothesis that tumor grade and an indolent clinical course predict improved survival after TPN administration, we examined data for the 33 patients with known tumor grade. The median time to death in patients with low-grade tumors versus high-grade tumors was 8 and 4 months, respectively (P = 0.16). Similarly, patients diagnosed with metastatic disease within the last year versus those diagnosed earlier had no significant difference in survival, i.e., 5 and 7 months, respectively, (P = 0.74). To test our second hypothesis that prominent cancer symptoms predict a poorer outcome, we explored differences in survival in patients with symptoms at the time of TPN initiation versus those without symptoms and observed no significant differences in survival. The median survival periods were 5 and 6 months, respectively (P = 0.31). Finally, we explored the hypothesis that cancer therapy after the initiation of TPN was associated with an improved survival. Again, we did not find a statistically significant difference in median survival between groups. Patients who received cancer therapy versus those who did not had median survivals of 14 and 4 months, respectively (P = 0.17).

TPN-related complications included 18 catheter infections (1 per 2.8 TPN-years), 4 central venous thromboses (1 per 12.6 TPN-years), 3 pneumothoraces (1 per 16.7 TPN-years), and 2 episodes of TPN-related liver disease (1 per 25.1 TPN-years).


The current study retrospectively reviewed a single institution's 20-year experience with home TPN in patients with incurable, metastatic disease. The study included a relatively small group of 52 patients, < 15% of all patients who had received TPN at our institution during this interval. Although this patient group is limited in size, this small percentage suggests that the use of TPN at our institution under these circumstances is quite conservative and, for the most part, in keeping with general practice guidelines.12, 13 Nevertheless, to our knowledge, this 52-patient study represents the largest retrospective series comprising patients with metastatic, incurable cancer who received TPN. Although other series may be slightly larger,7 not all patients in these other series clearly had incurable malignant diseases. Hence, the current study offers an opportunity to make several important points.

One noteworthy point is that some patients who receive TPN, even in the setting of metastatic, incurable cancer, can live well beyond a few months. Fifteen patients lived ≥ 1 year. One patient is still alive at the time of the current study, and is still receiving TPN 12 years later. Because cessation of energy intake is associated with a life expectancy of only a few months,11 this patient and the other long-term survivors in our series suggest that TPN can be a lifesaving and life-sustaining intervention for some patients with incurable cancer. TPN-related complication rates were acceptable and were somewhat lower in the current study compared with previous ones, including a study from our own institution.14 Thus, it appears that a small group of patients in our study derived a net benefit from TPN.

Another noteworthy point is that the patients who received TPN in the outpatient setting represent an unusual group of patients with cancer, not typical of the majority of patients with cancer. First, many had rare tumor histologies, such as carcinoid/islet cell tumors or amyloidosis, the former being the most common cancer type. Patients with the more common malignancies, such as lung, colon, or breast carcinoma were poorly represented within this group of patients. Second, 83% of patients already had lost weight before the initiation of TPN, but the etiology of this weight loss was related to mechanical gastrointestinal problems, such as obstruction or malabsorption. In contrast, anorexia is often cited as a primary cause of weight loss in patients with advanced-stage cancer.15 However, of these 52 patients, only 2 received TPN because of anorexia. Thus, even after initial inspection, this group of patients is not representative of the majority of patients with incurable, metastatic disease, an observation that invites some caution in extrapolating findings from the current study to all other patients with cancer.

Moreover, none of the potentially predictive factors was significantly associated with improved survival. Why were we unable to predict who would survive longer? First, the small sample of only 52 patients may not have allowed adequate power to detect subtle associations, although one might question the clinical utility of a prognostic effect that is not manifested among 52 patients. Second, these 52 patients represent a culled group. In effect, all patients received TPN, presumably after a great deal of deliberation on the part of the physicians involved in their care. Had we included a more diverse group of cancer patients, who were more representative of the majority of patients with incurable, metastatic disease, we might have detected strong positive associations between survival and these hypothesized predictive factors. Thus, we conclude that a patient should not be deprived of TPN if multidisciplinary clinical deliberation and clear discussions with patients and family members all suggest that a particular patient with cancer might potentially benefit. Based on the results of the current study, we cannot conclude that any specific clinical factor can substitute for a thoughtful, judicious assessment undertaken on a patient-by-patient basis.

Of parenthetical note, we did not evaluate long-term survivorship based on type of malignancy. The multitude of different tumor types in this series, many of which were represented by fewer than a handful of patients, did not allow for such meaningful analyses. It is true that many of the patients who survived ≥ 1 year had malignancies that are perceived as relatively indolent, for example, carcinoid/islet cell tumors. However, one-half of the patients with these carcinoid/islet cell tumors did not live 1 year. In addition, 1 patient with a high-grade ovarian tumor lived for almost 3 years, and another with a high-grade adenocarcinoma of the pancreas lived for > 2 years. These observations suggest that type of malignancy alone should not dictate whether a patient with cancer in this setting receives home TPN.

Finally, the retrospective nature of the current study precluded quality of life and cost assessment. For some patients, TPN may have improved their quality of life. For others, however, especially for those in the postoperative setting where surgical drains and open wounds required constant, ongoing care, TPN at home may have brought with it yet another undue burden. In addition, the cost of TPN is difficult to estimate from a 20-year retrospective study because of fluctuations in medical costs and inflation rates. Previous studies have commented on the costs of this intervention,16 and there may be a role for examining this factor in future studies. Although the survival data presented suggest that the judicious use of TPN can benefit some patients, quality of life and cost data are necessary to solidify that conclusion.

In summary, the current literature states indefatigably that the majority of patients with incurable, metastatic disease should not receive TPN at home. However, we, as well as others,6–10 found that a subgroup of patients who have advanced-stage, incurable cancer and intestinal failure benefit from home TPN and find it life sustaining. There is a clear need to provide a balanced in-depth discussion with the patient and family members before initiating TPN and to emphasize that only a small group of studies and case reports justify its use. We emphasize the need for conducting robust studies in the future to identify which patients with incurable metastatic disease would gain long-term benefits from home TPN.