. 2020 May 4;74(11):1519–1535. doi: 10.1038/s41430-020-0629-0

Table 1.

Characteristics of included studies.

Prospective cohort studies	Age, years	Sample size, gender	Cancer diagnoses and treatment	Malnutrition proxy marker(s) and units	Patient outcome(s)	Follow up	Study results [95% CI]	Quality score
Aaldricks, The Netherlands [31]	≥70 YO Mean 76 ± 4.8 Range 70–88	n = 55 F = 53 M = 2	Advanced breast cancer. Chemotherapy.	Alb (</≥ 35 g/l) Hb (</≥7.5 mmol/l)	Overall mortality	Median 16 ± 13.7 months	No association between proxy markers and outcomes.	8.5/10 Risk of selection bias.
≥4 Vs <4 cycles of chemotherapy	Median 11 months Range 0–57.
Aaldricks, The Netherlands [32]	≥70 YO Mean 77	n = 44 F = 25 M = 19	Non-Hodgkin’s Lymphoma; diffuse large-β cell lymphoma and follicular lymphoma grade III. R-CHOP treatment.	Alb (</≥35 g/l). Hb (6.8 mmol/l).	Completion of chemotherapy. Mortality.	Median 46 months (0–101).	Hb associated with early treatment withdrawal: multivariate OR 5.41 [0.99–29.8] p = 0.05 and mortality: HR 4.90 [1.76–13.7], p = 0.0002.	9.5/10
Aaldricks, The Netherlands [33]	≥70 YO Median 75 Range 70–92	n = 494 F = 248 M = 246	Various cancer diagnoses. Chemotherapy.	Declining food intake 3/12 (severe or moderate decrease/no decrease). Weight loss 3/12.	Feasibility of Chemotherapy ≥4 Vs <4 cycles	Median 17 months Range 1–101.	Declining food intake OR 2.00 [1.34–3.00], weight loss 3/12 OR 1.88 [1.26–2.80] associated with feasibility of chemotherapy in univariate analysis. Declining food intake, fluid intake ≤5 cups/day OR 1.76 [1.23–2.52] and weight loss 6/12 OR 1.38 [1.13–1.69] associated with mortality in univariate analysis.	7.0/10 Missing data, unclear recruitment.
Declining food intake 3/12 Reduced fluid intake (≤3/3–5/≥5 cups per day). Unintentional weight loss (3 kg in 1/12 or 6 kg in 2/12).	Overall mortality
Baitar, Belgium [34]	≥70 YO Median 77 Range 70–95	n = 328 F = 194 M = 134	Breast (38.4%, Colorectal 35.4%, Lung 15.5%, Prostate 6.4%, Ovarian 4.3%). 63.7% new diagnosis, 36.3% progression or recurrence. Surgery, chemotherapy, radiotherapy hormonal therapy.	Hb (</≥11.8/12 g/dl)* Alb </≥ 35/37 g/l)* CRP </>5/≥5 mg/l* Markers also analysed as continuous variables	Overall survival	Median 60.3 months [95% CI: 58.6–62.6].	Hb, CRP and Alb associated with outcome as dichotomous variables: Hb HR 1.51 [1.16–1.96]. Alb HR 2.91 [1.44–2.52]. CRP: HR 1.82 [1.37–2.43]. CRP associated with outcome as continuous variable: HR 1.08 [1.06–1.11].	7.5/10 Unclear recruitment method.
Bourdel-Marchasson, France [35]	≥70 YO	n = 606 F = 287 M = 319	Lung, Colon, Stomach, Pancreas, Ovary, Bladder, CUPSs, Biliary duct, Breast. Life expectancy ≥12 weeks. First line chemotherapy.	% Weight loss (none/<5%, 5–10%, >10%, missing). Decreased food intake 3/12 (severe/moderate/no decrease). Actual weight loss 3/12 (>3 kg, 1–3 kg/unknown/none). BMI (<19/>19–<21/>21–<23/>23 kg/m²) Daily full meals (1/2/3 meals). Protein-rich foods (low/intermediate/high). Fruit & vegetable intake (<2/≥2 servings/day). Fluid intake (<3/3–5/>5 cups/day). Self-view of nutritional status (malnourished/uncertain/no problem). Mode of feeding (assistance/self-fed with difficulty/no problem). MAC (<21/<21–<22/>22 cm). Calf circumference (<31/>31 cm).	1 year mortality	12 months.	In univariate analysis; reduced food intake 3.12, weight loss >3 kg or unknown weight loss, BMI <23, number of full meals per day, <2 servings fruit and vegetables/day, self-fed with some difficulty, self-view of nutritional status, mid-arm circumference <21 cm, calf circumference <31 cm associated with outcome.	5.5/10 n = 33 lost to follow-up, unclear recruitment method, risk of bias in data collection.
Chaufour-André, France [36]	≥70 YO	n = 71 F = 33 M = 38	Digestive, Upper aero- digestive, Gynaecological, Lung, Sarcomas, Other. Surgery for neoplastic pathology.	NRI </> 97.5. Unintentional weight loss.	Major complications. Infectious complications. Post-operative confusion.	1 month post-discharge.	Univariate analysis; NRI associated with post-operative complications: OR 0.79 [0.66–0.95]. No risk factors for postoperative complications could be identified.	6.5/10 Confounding not accounted for, risk of bias in recruitment.
Extermann, USA [37]	≥70 YO Median 75.5 Range 70–92	n = 518 F = 261 M = 257	Lung, Breast, Non-Hodgkin’s Lymphoma, Colorectal, Bladder, other. Chemotherapy.	BMI > 25 kg/m². Hb (g/dl). Alb (g/dl).	Chemotherapy toxicity; grade 4 haematological or grade 3/4 non-haematological.	6 months.	No association between proxy markers and outcome.	6.0/10 Risk of bias recording proxy markers and outcomes.
Hoppe, France [38]	≥70 YO Median 77.4 Range 70–93	n = 299 F = 122 M = 177	Colon, Pancreatic, Stomach, Ovarian, Bladder, Prostate, Lung, Non-Hodgkin’s lymphoma, CUPs. First-line chemotherapy.	Weight loss (</≥ 10%). BMI (<19/19–23/≥23 kg/m²). Alb (</≥ 35 g/l). CRP (</≥ 5 mg/l).	Functional decline (ADL score).	After first cycle of chemotherapy	Weight loss associated with functional decline in univariate analysis OR 1.86 [no CIs] p = 0.05. No multivariate analysis given.	6.5/10 Risk of bias in recruitment, inappropriate follow up time.
Hsu, Canada [39]	≥70 YO Median 74.5 Range 70–84	n = 24 F = 7 M = 17	Colorectal or Thoracic cancer. Chemotherapy.	Hand-grip strength (bottom 20th percentile)	Chemotherapy toxicity (grade 3–5). Dose reduction or delay due to chemotherapy toxicity. Discontinuation of chemotherapy due to toxicity. Hospitalisation or ED visit due to chemotherapy.	12 months.	p values only, no association between proxy marker and outcomes.	2.0/10 Risk of confounding, unclear recruitment, inappropriate conclusions.
Kaibori, Japan [40]	≥70 YO Median 77 Mean 78.2 ± 4.8 Range 70–89	n = 71 F = 19 M = 52	Hepatocellular carcinoma. Hepatic resection.	BMI (</≥ 22 kg/m²). Alb (</≥4 g/dl).	Post-operative complications (Clavien-Dindo grade 2–4b)	Length of hospital stay: 13 days (6–189).	Alb associated with outcome in univariate analysis OR 3.66 [1.14–1.76], p = 0.00292.	7.5/10 Risk of selection bias in recruitment and inclusion criteria.
Kanesvaran, Singapore [41]	≥70 YO Median 77 Range 70–94	n = 249 F = 96 M = 153	All cancer diagnoses; Lung, Colorectal and Genitourinary 83.5%	BMI (</≥ 30 kg/m²). Hb (</≥ 12 g/dl). Alb (</≥35 g/l).	Survival (median months)	No info.	Hb and albumin associated with outcome in univariate analysis. Multivariate analysis for Hb not given.	9.0/10 Missing data.
Kim, South Korea [42]	≥70 YO	n = 301 F = 93 M = 208	Colorectal, Lung, Hepato-biliary, Stomach, Other. Stage III, IV or unknown. First-line chemotherapy.	Daily fluid intake (</> 3 cups per day).	≥Grade 3 chemotherapy toxicity	Post-chemotherapy cycles (range 25–75% 2–7 cycles).	Daily fluid intake associated with outcome.	7.5/10 Recruitment method not described.
Lu, China [43]	≥80 YO Range 80–93	n = 165 F = 30 M = 132	Gastric cancer. Surgical resection.	PNI </≥45.	Systematic complications. Local complications. Overall survival. Recurrence free survival. Cancer specific survival.	5 years.	PNI associated with recurrence-free survival	9.5/10
Marenco, Italy [44]	≥70 YO Median 78 Mean 78 ± 4.8	n = 571 F = 220 M = 351	Colorectal, Gastro-intestinal, Renal, Bladder, Other.	BMI </≥21 kg/m²	Treatment recommendation (active vs palliative care). Survival.	Up to 60 months.	BMI associated with outcome.	6.5/10 High risk of selection bias.

Prospective cohort studies

Age, years

Sample size, gender

Cancer diagnoses and treatment

Malnutrition proxy marker(s) and units

Patient outcome(s)

Follow up

Study results [95% CI]

Quality score

Aaldricks,

The Netherlands [31]

≥70 YO

Mean 76 ± 4.8

Range 70–88

n = 55

F = 53

M = 2

Advanced breast cancer. Chemotherapy.

Alb (</≥ 35 g/l)

Hb (</≥7.5 mmol/l)

Overall mortality

Median 16 ± 13.7 months

No association between proxy markers and outcomes.

8.5/10

Risk of selection bias.

≥4 Vs <4 cycles of chemotherapy

Median 11 months

Range 0–57.

Aaldricks,

The Netherlands [32]

≥70 YO

Mean 77

n = 44

F = 25

M = 19

Non-Hodgkin’s Lymphoma; diffuse large-β cell lymphoma and follicular lymphoma grade III.

R-CHOP treatment.

Alb (</≥35 g/l).

Hb (6.8 mmol/l).

Completion of chemotherapy.

Mortality.

Median 46 months (0–101).

Hb associated with early treatment withdrawal: multivariate OR 5.41 [0.99–29.8] p = 0.05 and mortality: HR 4.90 [1.76–13.7], p = 0.0002.

9.5/10

Aaldricks,

The Netherlands [33]

≥70 YO

Median 75

Range 70–92

n = 494

F = 248

M = 246

Various cancer diagnoses.

Chemotherapy.

Declining food intake 3/12 (severe or moderate decrease/no decrease).

Weight loss 3/12.

Feasibility of Chemotherapy

≥4 Vs <4 cycles

Median 17 months

Range 1–101.

Declining food intake OR 2.00 [1.34–3.00], weight loss 3/12 OR 1.88 [1.26–2.80] associated with feasibility of chemotherapy in univariate analysis.

Declining food intake, fluid intake ≤5 cups/day OR 1.76 [1.23–2.52] and weight loss 6/12 OR 1.38 [1.13–1.69] associated with mortality in univariate analysis.

7.0/10

Missing data, unclear recruitment.

Declining food intake 3/12

Reduced fluid intake (≤3/3–5/≥5 cups per day).

Unintentional weight loss (3 kg in 1/12 or 6 kg in 2/12).

Overall mortality

Baitar,

Belgium [34]

≥70 YO

Median 77

Range 70–95

n = 328

F = 194

M = 134

Breast (38.4%, Colorectal 35.4%, Lung 15.5%, Prostate 6.4%, Ovarian 4.3%).

63.7% new diagnosis, 36.3% progression or recurrence.

Surgery, chemotherapy, radiotherapy hormonal therapy.

Hb (</≥11.8/12 g/dl)*

Alb </≥ 35/37 g/l)*

CRP </>5/≥5 mg/l*

Markers also analysed as continuous variables

Overall survival

Median 60.3 months

[95% CI: 58.6–62.6].

Hb, CRP and Alb associated with outcome as dichotomous variables: Hb HR 1.51 [1.16–1.96].

Alb HR 2.91 [1.44–2.52].

CRP: HR 1.82 [1.37–2.43]. CRP associated with outcome as continuous variable: HR 1.08 [1.06–1.11].

7.5/10

Unclear recruitment method.

Bourdel-Marchasson,

France [35]

≥70 YO

n = 606

F = 287

M = 319

Lung, Colon, Stomach, Pancreas, Ovary, Bladder, CUPSs, Biliary duct, Breast. Life expectancy ≥12 weeks.

First line chemotherapy.

% Weight loss (none/<5%, 5–10%, >10%, missing).

Decreased food intake 3/12 (severe/moderate/no decrease).

Actual weight loss 3/12 (>3 kg, 1–3 kg/unknown/none).

BMI (<19/>19–<21/>21–<23/>23 kg/m²)

Daily full meals (1/2/3 meals).

Protein-rich foods (low/intermediate/high).

Fruit & vegetable intake (<2/≥2 servings/day).

Fluid intake (<3/3–5/>5 cups/day).

Self-view of nutritional status (malnourished/uncertain/no problem).

Mode of feeding (assistance/self-fed with difficulty/no problem).

MAC (<21/<21–<22/>22 cm).

Calf circumference (<31/>31 cm).

1 year mortality

12 months.

In univariate analysis; reduced food intake 3.12, weight loss >3 kg or unknown weight loss, BMI <23, number of full meals per day, <2 servings fruit and vegetables/day, self-fed with some difficulty, self-view of nutritional status, mid-arm circumference <21 cm, calf circumference <31 cm associated with outcome.

5.5/10

n = 33 lost to follow-up, unclear recruitment method, risk of bias in data collection.

Chaufour-André,

France [36]

≥70 YO

n = 71

F = 33

M = 38

Digestive, Upper aero- digestive, Gynaecological, Lung, Sarcomas, Other. Surgery for neoplastic pathology.

NRI </> 97.5.

Unintentional weight loss.

Major complications. Infectious complications.

Post-operative confusion.

1 month post-discharge.

Univariate analysis; NRI associated with post-operative complications: OR 0.79 [0.66–0.95]. No risk factors for postoperative complications could be identified.

6.5/10

Confounding not accounted for, risk of bias in recruitment.

Extermann,

USA [37]

≥70 YO

Median 75.5

Range 70–92

n = 518

F = 261

M = 257

Lung, Breast, Non-Hodgkin’s Lymphoma, Colorectal, Bladder, other.

Chemotherapy.

BMI > 25 kg/m².

Hb (g/dl).

Alb (g/dl).

Chemotherapy toxicity; grade 4 haematological or grade 3/4 non-haematological.

6 months.

No association between proxy markers and outcome.

6.0/10

Risk of bias recording proxy markers and outcomes.

Hoppe,

France [38]

≥70 YO

Median 77.4

Range 70–93

n = 299

F = 122

M = 177

Colon, Pancreatic, Stomach, Ovarian, Bladder, Prostate, Lung, Non-Hodgkin’s lymphoma, CUPs.

First-line chemotherapy.

Weight loss (</≥ 10%).

BMI (<19/19–23/≥23 kg/m²).

Alb (</≥ 35 g/l).

CRP (</≥ 5 mg/l).

Functional decline (ADL score).

After first cycle of chemotherapy

Weight loss associated with functional decline in univariate analysis OR 1.86 [no CIs] p = 0.05. No multivariate analysis given.

6.5/10

Risk of bias in recruitment, inappropriate follow up time.

Hsu,

Canada [39]

≥70 YO

Median 74.5

Range 70–84

n = 24

F = 7

M = 17

Colorectal or Thoracic cancer.

Chemotherapy.

Hand-grip strength (bottom 20th percentile)

Chemotherapy toxicity (grade 3–5).

Dose reduction or delay due to chemotherapy toxicity.

Discontinuation of chemotherapy due to toxicity.

Hospitalisation or ED visit due to chemotherapy.

12 months.

p values only, no association between proxy marker and outcomes.

2.0/10

Risk of confounding, unclear recruitment, inappropriate conclusions.

Kaibori,

Japan [40]

≥70 YO

Median 77 Mean 78.2 ± 4.8

Range 70–89

n = 71

F = 19

M = 52

Hepatocellular carcinoma.

Hepatic resection.

BMI (</≥ 22 kg/m²).

Alb (</≥4 g/dl).

Post-operative complications (Clavien-Dindo grade 2–4b)

Length of hospital stay: 13 days (6–189).

Alb associated with outcome in univariate analysis OR 3.66 [1.14–1.76], p = 0.00292.

7.5/10

Risk of selection bias in recruitment and inclusion criteria.

Kanesvaran,

Singapore [41]

≥70 YO

Median 77

Range 70–94

n = 249

F = 96

M = 153

All cancer diagnoses; Lung, Colorectal and Genitourinary 83.5%

BMI (</≥ 30 kg/m²).

Hb (</≥ 12 g/dl).

Alb (</≥35 g/l).

Survival (median months)

No info.

Hb and albumin associated with outcome in univariate analysis. Multivariate analysis for Hb not given.

9.0/10

Missing data.

Kim,

South Korea [42]

≥70 YO

n = 301

F = 93

M = 208

Colorectal, Lung, Hepato-biliary, Stomach, Other. Stage III, IV or unknown.

First-line chemotherapy.

Daily fluid intake (</> 3 cups per day).

≥Grade 3 chemotherapy toxicity

Post-chemotherapy cycles (range 25–75% 2–7 cycles).

Daily fluid intake associated with outcome.

7.5/10

Recruitment method not described.

Lu,

China [43]

≥80 YO

Range 80–93

n = 165

F = 30

M = 132

Gastric cancer.

Surgical resection.

PNI </≥45.

Systematic complications. Local complications.

Overall survival. Recurrence free survival.

Cancer specific survival.

5 years.

PNI associated with recurrence-free survival

9.5/10

Marenco,

Italy [44]

≥70 YO

Median 78

Mean 78 ± 4.8

n = 571

F = 220

M = 351

Colorectal, Gastro-intestinal, Renal, Bladder, Other.

BMI </≥21 kg/m²

Treatment recommendation (active vs palliative care).

Survival.

Up to 60 months.

BMI associated with outcome.

6.5/10

High risk of selection bias.

Retrospective cohort studies	Age, years	Sample size, gender	Cancer diagnoses and treatment	Malnutrition proxy markers	Patient outcomes	Follow up	Study results [95% CI]	Quality score
Fiorelli, Italy [45]	≥70 YO Median 75 Mean 74.9 ± 2.6 Range 71–93	n = 117 F = 23 M = 94	Non-small cell lung cancer. Curative resection.	BMI (≤/> 18.5 kg/m²) Alb (≥35 g/l) Weight loss (≥5% 3/12)	Major complications. Early death (<3/12 post procedure).	3 months.	BMI and albumin associated with major complications in univariate analysis.	3.0/10 Risk of selection/recruitment bias, risk of bias in data collection, statistics errors.
Harimoto, Japan [46]	≥70 YO	n = 139 F = 41 M = 98	Hepatocellular carcinoma. Curative hepatic resection.	BMI (kg/m²). Alb (g/dl). CRP (mg/dl). PNI.	Overall survival. Disease-free survival.	No info.	Univariate analysis; CRP associated with disease-free survival: HR 1.35 [1.14–1.59].	6.0/10 Risk of bias in data collection.
Kim, South Korea [47]	≥70 YO Median 76 IQR 72–80	n = 122 F = 37 M = 85	Primary non-small cell lung cancer, ≥ stage IIIB. Admitted to hospital.	BMI (<18 kg/m²).	Survival	6.2 months (IQR: 2.5–15.3).	BMI associated with outcome.	8.0/10 Missing data. Risk of bias in data collection.
Kushiyama, Japan [48]	≥75 YO Mean 79.6 ± 3.8	n = 348 F = 118 M = 230	Gastric cancer. Gastrectomy.	BMI (<22 kg/m²). GNRI (<92).	Post-operative complications (Clavien-Dindo grade 2–4)	No info.	GNRI associated with outcome	9.0/10
Lai, Canada [49]	≥80YO Median 83 Range 80–92	n = 60 F = 29 M = 31	Metastatic colorectal cancer. Chemotherapy.	Hb </≥100 g/l.	Chemotherapy dose reduction/omission or delay >1 week. Chemotherapy discontinuation due to toxicity. Hospitalisation within 30 days of chemotherapy. Overall survival.	No info.	Hb associated with overall survival	5.5/10 Recruitment not discussed. Missing data.
Mikami, Japan [50]	≥70 YO	n = 267 F = 92 M = 175	Primary gastric cancer. Curative gastrectomy.	BMI (kg/m²). Hb (g/dl). PNI </≥ 40.	Overall survival. Gastric cancer specific survival.	5 years.	BMI and PNI associated with overall survival.	5.5/10 Risk selection bias.
Mosk, The Netherlands [51]	≥70 YO Median 76 IQR: 73–80	n = 251 F = 110 M = 141	Colorectal cancer. Elective surgery.	Low skeletal muscle mass (<35.17 females cm²/m², <43.19 cm²/m² males). Low skeletal muscle density.	Post-operative delirium.	Length of hospital stay.	Low skeletal muscle mass associated with outcome.	9.5/10
Neuman, USA [52]	≥80 YO Mean 84.4 ± 3.7	n = 12979 F = 7976 M = 5003	Colon cancer. Surgical resection.	Weight loss.	90 day mortality. 1 year mortality.	1 year.	No association between proxy marker and outcomes.	7.5/10 Risk of bias in data collection.
Sakurai, Japan [53]	≥75 YO Mean 79 ± 3.4	n = 147 F = 52 M = 95	Gastric cancer. Curative gastrectomy.	BMI </≥ 22 kg/m². PNI ≤/> 43.8.	Overall survival.	5 years.	PNI associated with outcome.	9.0/10 Recruitment method not discussed.
Sakurai, Japan [54]	≥75 YO Mean 79.2 ± 3.5	n = 175 F = 59 M = 116	Gastric cancer, stage 1. Gastrectomy.	BMI (<22 kg/m²). PNI (<45).	5 year overall survival.	5 years.	PNI associated with outcome.	7.5/10 Risk of bias handing missing data and data collection.
Sekiguchi, Japan [55]	≥85 YO Median 86 Range 85–93	n = 108 F = 26 M = 82	Gastric cancer. Endoscopic submucosal dissection.	PNI </≥ 44.6. BMI </≥ 24.3 kg/m²	Overall survival.	5 years.	PNI associated with outcome.	6.5/10 Risk of bias handing missing data, data presentation.
Shoji, Japan [56]	≥75 YO Median 78 Range 75–91	n = 272 F = 117 M = 155	Primary lung cancer. Surgical resection.	Preoperative BMI </≥18.5 kg/m² Preoperative PNI ≤/> 49.6. Preoperative CONUT </≥1. Preoperative GNRI ≤/>98.	Post-operative comorbidities. Overall survival.	Median 51 months Range 0–132.	GNRI associated with outcome.	8.0/10 Risk of bias in data collection.
Stangl-Kremser, Austria [57]	≥70 YO Median 82 IQR 75–86	n = 68 F = 13 M = 55	Urothelial carcinoma of the bladder. Transurethral resection.	PNI </≥45.2. CONUT. BMI kg/m²	Overall survival. Cancer specific survival.	Median 12.5 months (IQR: 5.1–23.5).	PNI associated with overall survival.	3.0/10 Risk of confounding, missing data.
Takama, Japan [58]	≥75 YO	n = 190 F = 60 M = 130	Gastric cancer. Gastrectomy.	Alb </≥3.5 g/dl. PNI </≥40.	Complications (Clavien-Dindo Grade ≥2).	Mean 46 months.	PNI p = 0.005 [no CI] and Alb p = 0.019 [no CI] associated with complications in ages ≥85.	4.0/10 Recruitment method not discussed. Data presentation.
Tei, Japan [59]	≥71 YO	n = 129 F = 54 M = 75	Colorectal cancer. Surgery.	PNI (comparison of means).	Post-operative delirium.	30 days post-surgery.	PNI associated with outcome.	7.0/10 Recruitment method not discussed.
Tei, Japan [60]	≥75 YO Median 79 Range 75–93	n = 311 F = 140 M = 171	Colorectal cancer. Laparoscopic surgery.	PNI (comparison of means). Hb (10 g/dl).	Post-operative delirium.	30 days post-surgery.	No association between proxy markers and outcome.	8.0/10 Risk of selection bias and bias in data collection.
Tominaga, Japan [61]	≥70 YO	n = 239 F = 118 M = 121	Colorectal cancer. Curative resection.	PNI. Body weight. BMI. Alb. Hb (10–13/13–16/16–18/<10/> 18 g/dl).	Post-operative complications (Clavien-Dindo grade 2–5).	Median 25.7 months (range 0.2–69.2).	PNI p ≤ 0.05 [no CI] and Alb p = 0.04 [no CI] associated with complications.	3.0/10 Risk of bias in data collection and data presentation. Missing data.
Toya, Japan [62]	≥75 YO Median 78 Range 75–88	n = 87 F = 22 M = 65	Non-curative gastric cancer. Endoscopic submucosal dissection.	PNI </≥ 44.8. GNRI ≤/> 92.	Overall survival.	Median 6.7 years (range 0.1–14.8).	No association between proxy markers and outcome.	8.0/10 Risk of selection bias and data collection
Ueno, Japan [63]	≥75 YO Median Range 75–91	n = 117 F = 35 M =82	Gastric cancer. Curative surgery.	PNI </≥ 40.	Overall survival. Disease-specific survival.	Median 52.9 (range 1.0–117.5).	No association between proxy marker and outcomes.	6.5/10 Risk of bias in data collection, missing data.
Watanabe, Japan [64]	≥75 YO Median Range	n = 99 F = 23 M = 76	Gastric cancer. Curative intent gastrectomy.	PNI </≥44.7.	Overall survival.	5 years.	Proxy marker associated with outcome.	9.0/10
Watanabe, Japan [65]	≥75 YO Median 79 Range 75–88	n = 131 F = 63 M = 68	Primary lung cancer. Complete surgical resection.	PNI </≥ 45.	Overall survival.	5 years.	Proxy marker associated with outcome.	9.0/10 Risk of selection bias
Yoshimatsu, Japan [66]	≥80 YO Median 83 Range 80–90	n = 76 F = 40 M = 36	Colorectal cancer. Curative resection.	PNI </≥40.	3 and 5 year survival.	Median 30 months.	No association between proxy markers and outcomes.	2.0/10 Risk of bias data collection, confounding, selection bias, data presentation.
Zauderer, USA [67]	≥70 YO Median 75 Range 70–92	n = 70 F = 20 M = 50	Metastatic non-small cell lung cancer. Chemotherapy.	Unintentional weight loss (Y/N). Alb </≥3.5 g/dl. Anaemia (Y/N).	Chemotherapy complications; grade 3/4 haematologic and grade 4 non-haematologic toxicity. Treatment delay. Dose reduction. Hospitalisation. Discontinuation of chemotherapy due to toxicity.	No info.	No association between proxy markers and outcomes.	1.5/10 Confounding not accounted for. Convenience sample. Risk of bias in data collection. Data presentation.
Zhou, China [68]	≥70 YO Median 79 Range 75–91	n = 164 F = 67 M = 97	Oesophageal cancer. Radiotherapy ± chemotherapy.	NRI </≥100.	2 year overall survival. 2 year local-regional failure-free survival. 2 year distance metastasis-free survival.	2 years.	Proxy marker associated with outcomes.	8.0/10 Risk of bias in data collection. Missing data.

Retrospective cohort studies

Age, years

Sample size, gender

Cancer diagnoses and treatment

Malnutrition proxy markers

Patient outcomes

Follow up

Study results [95% CI]

Quality score

Fiorelli,

Italy [45]

≥70 YO

Median 75

Mean 74.9 ± 2.6

Range 71–93

n = 117

F = 23

M = 94

Non-small cell lung cancer.

Curative resection.

BMI (≤/> 18.5 kg/m²)

Alb (≥35 g/l)

Weight loss (≥5% 3/12)

Major complications.

Early death (<3/12 post procedure).

3 months.

BMI and albumin associated with major complications in univariate analysis.

3.0/10

Risk of selection/recruitment bias, risk of bias in data collection, statistics errors.

Harimoto,

Japan [46]

≥70 YO

n = 139

F = 41

M = 98

Hepatocellular carcinoma.

Curative hepatic resection.

BMI (kg/m²).

Alb (g/dl).

CRP (mg/dl).

PNI.

Overall survival.

Disease-free survival.

No info.

Univariate analysis; CRP associated with disease-free survival: HR 1.35 [1.14–1.59].

6.0/10

Risk of bias in data collection.

Kim,

South Korea [47]

≥70 YO

Median 76

IQR 72–80

n = 122

F = 37

M = 85

Primary non-small cell lung cancer, ≥ stage IIIB.

Admitted to hospital.

BMI (<18 kg/m²).

Survival

6.2 months (IQR: 2.5–15.3).

BMI associated with outcome.

8.0/10

Missing data. Risk of bias in data collection.

Kushiyama,

Japan [48]

≥75 YO

Mean 79.6 ± 3.8

n = 348

F = 118

M = 230

Gastric cancer.

Gastrectomy.

BMI (<22 kg/m²).

GNRI (<92).

Post-operative complications (Clavien-Dindo grade 2–4)

No info.

GNRI associated with outcome

9.0/10

Lai,

Canada [49]

≥80YO

Median 83

Range 80–92

n = 60

F = 29

M = 31

Metastatic colorectal cancer.

Chemotherapy.

Hb </≥100 g/l.

Chemotherapy dose reduction/omission or delay >1 week.

Chemotherapy discontinuation due to toxicity.

Hospitalisation within 30 days of chemotherapy.

Overall survival.

No info.

Hb associated with overall survival

5.5/10

Recruitment not discussed. Missing data.

Mikami, Japan [50]

≥70 YO

n = 267

F = 92

M = 175

Primary gastric cancer.

Curative gastrectomy.

BMI (kg/m²).

Hb (g/dl).

PNI </≥ 40.

Overall survival.

Gastric cancer specific survival.

5 years.

BMI and PNI associated with overall survival.

5.5/10

Risk selection bias.

Mosk,

The Netherlands [51]

≥70 YO

Median 76

IQR: 73–80

n = 251

F = 110

M = 141

Colorectal cancer.

Elective surgery.

Low skeletal muscle mass (<35.17 females cm²/m², <43.19 cm²/m² males).

Low skeletal muscle density.

Post-operative delirium.

Length of hospital stay.

Low skeletal muscle mass associated with outcome.

9.5/10

Neuman,

USA [52]

≥80 YO

Mean 84.4 ± 3.7

n = 12979

F = 7976

M = 5003

Colon cancer.

Surgical resection.

Weight loss.

90 day mortality.

1 year mortality.

1 year.

No association between proxy marker and outcomes.

7.5/10

Risk of bias in data collection.

Sakurai,

Japan [53]

≥75 YO

Mean 79 ± 3.4

n = 147

F = 52

M = 95

Gastric cancer.

Curative gastrectomy.

BMI </≥ 22 kg/m².

PNI ≤/> 43.8.

Overall survival.

5 years.

PNI associated with outcome.

9.0/10

Recruitment method not discussed.

Sakurai,

Japan [54]

≥75 YO

Mean 79.2 ± 3.5

n = 175

F = 59

M = 116

Gastric cancer, stage 1.

Gastrectomy.

BMI (<22 kg/m²).

PNI (<45).

5 year overall survival.

5 years.

PNI associated with outcome.

7.5/10

Risk of bias handing missing data and data collection.

Sekiguchi,

Japan [55]

≥85 YO

Median 86

Range 85–93

n = 108

F = 26

M = 82

Gastric cancer.

Endoscopic submucosal dissection.

PNI </≥ 44.6.

BMI </≥ 24.3 kg/m²

Overall survival.

5 years.

PNI associated with outcome.

6.5/10

Risk of bias handing missing data, data presentation.

Shoji,

Japan [56]

≥75 YO

Median 78

Range 75–91

n = 272

F = 117

M = 155

Primary lung cancer.

Surgical resection.

Preoperative BMI </≥18.5 kg/m²

Preoperative PNI ≤/> 49.6.

Preoperative CONUT </≥1.

Preoperative GNRI ≤/>98.

Post-operative comorbidities.

Overall survival.

Median 51 months

Range 0–132.

GNRI associated with outcome.

8.0/10

Risk of bias in data collection.

Stangl-Kremser,

Austria [57]

≥70 YO

Median 82

IQR 75–86

n = 68

F = 13

M = 55

Urothelial carcinoma of the bladder.

Transurethral resection.

PNI </≥45.2.

CONUT.

BMI kg/m²

Overall survival.

Cancer specific survival.

Median 12.5 months (IQR: 5.1–23.5).

PNI associated with overall survival.

3.0/10

Risk of confounding, missing data.

Takama,

Japan [58]

≥75 YO

n = 190

F = 60

M = 130

Gastric cancer.

Gastrectomy.

Alb </≥3.5 g/dl.

PNI </≥40.

Complications (Clavien-Dindo Grade ≥2).

Mean 46 months.

PNI p = 0.005 [no CI] and Alb p = 0.019 [no CI] associated with complications in ages ≥85.

4.0/10

Recruitment method not discussed. Data presentation.

Tei,

Japan [59]

≥71 YO

n = 129

F = 54

M = 75

Colorectal cancer.

Surgery.

PNI (comparison of means).

Post-operative delirium.

30 days post-surgery.

PNI associated with outcome.

7.0/10

Recruitment method not discussed.

Tei,

Japan [60]

≥75 YO

Median 79

Range 75–93

n = 311

F = 140

M = 171

Colorectal cancer.

Laparoscopic surgery.

PNI (comparison of means).

Hb (10 g/dl).

Post-operative delirium.

30 days post-surgery.

No association between proxy markers and outcome.

8.0/10

Risk of selection bias and bias in data collection.

Tominaga,

Japan [61]

≥70 YO

n = 239

F = 118

M = 121

Colorectal cancer.

Curative resection.

PNI.

Body weight.

BMI.

Alb.

Hb (10–13/13–16/16–18/<10/> 18 g/dl).

Post-operative complications (Clavien-Dindo grade 2–5).

Median 25.7 months (range 0.2–69.2).

PNI p ≤ 0.05 [no CI] and Alb p = 0.04 [no CI] associated with complications.

3.0/10

Risk of bias in data collection and data presentation. Missing data.

Toya,

Japan [62]

≥75 YO

Median 78

Range 75–88

n = 87

F = 22

M = 65

Non-curative gastric cancer.

Endoscopic submucosal dissection.

PNI </≥ 44.8.

GNRI ≤/> 92.

Overall survival.

Median 6.7 years (range 0.1–14.8).

No association between proxy markers and outcome.

8.0/10

Risk of selection bias and data collection

Ueno,

Japan [63]

≥75 YO

Median

Range 75–91

n = 117

F = 35

M =82

Gastric cancer.

Curative surgery.

PNI </≥ 40.

Overall survival.

Disease-specific survival.

Median 52.9 (range 1.0–117.5).

No association between proxy marker and outcomes.

6.5/10

Risk of bias in data collection, missing data.

Watanabe,

Japan [64]

≥75 YO

Median

Range

n = 99

F = 23

M = 76

Gastric cancer.

Curative intent gastrectomy.

PNI </≥44.7.

Overall survival.

5 years.

Proxy marker associated with outcome.

9.0/10

Watanabe,

Japan [65]

≥75 YO

Median 79

Range 75–88

n = 131

F = 63

M = 68

Primary lung cancer.

Complete surgical resection.

PNI </≥ 45.

Overall survival.

5 years.

Proxy marker associated with outcome.

9.0/10

Risk of selection bias

Yoshimatsu,

Japan [66]

≥80 YO

Median 83

Range 80–90

n = 76

F = 40

M = 36

Colorectal cancer.

Curative resection.

PNI </≥40.

3 and 5 year survival.

Median 30 months.

No association between proxy markers and outcomes.

2.0/10

Risk of bias data collection, confounding, selection bias, data presentation.

Zauderer,

USA [67]

≥70 YO Median 75

Range 70–92

n = 70

F = 20

M = 50

Metastatic non-small cell lung cancer.

Chemotherapy.

Unintentional weight loss (Y/N).

Alb </≥3.5 g/dl.

Anaemia (Y/N).

Chemotherapy complications; grade 3/4 haematologic and grade 4 non-haematologic toxicity.

Treatment delay. Dose reduction.

Hospitalisation.

Discontinuation of chemotherapy due to toxicity.

No info.

No association between proxy markers and outcomes.

1.5/10

Confounding not accounted for. Convenience sample. Risk of bias in data collection. Data presentation.

Zhou,

China [68]

≥70 YO

Median 79

Range 75–91

n = 164

F = 67

M = 97

Oesophageal cancer.

Radiotherapy ± chemotherapy.

NRI </≥100.

2 year overall survival.

2 year local-regional failure-free survival. 2 year distance metastasis-free survival.

2 years.

Proxy marker associated with outcomes.

8.0/10

Risk of bias in data collection. Missing data.

Cross-sectional studies	Age, years	Sample size, gender	Cancer diagnoses and treatment	Malnutrition proxy markers	Patient outcomes	Follow up	Study results [95% CI]	Quality score
Girre, France [69]	≥70 YO Median 79 Range 70–97	n = 105 F = 87 M = 18	Breast, Lung, Colorectal, Cervix, Endometrial, Ovarian, Prostate, Melanoma, Haematological. Other.	BMI (</≥ 23 kg/m²). Hb (</≥ 12 g/dl). Alb (20–35/>35 g/l).	Treatment plan modification.	NA	BMI associated with outcome, p = 0.029 [no CI].	3.5/10 Risk of bias in data collection, selection bias, Data presentation.
Rajasekaran, Singapore [70]	≥70 YO Median 77 Range 70–94	n = 244 F = 95 M = 149	Gastrointestinal, Lung, Genitourinary, Other.	BMI (</≥27.5 kg/m²). Hb (</≥ 12 g/dl). Dominant handgrip (per kg increase).	Caregiver burden.	NA	Hb associated with outcome	7.5/10 Risk of confounding, study design.

Cross-sectional studies

Age, years

Sample size, gender

Cancer diagnoses and treatment

Malnutrition proxy markers

Patient outcomes

Follow up

Study results [95% CI]

Quality score

Girre,

France [69]

≥70 YO

Median 79

Range 70–97

n = 105

F = 87

M = 18

Breast, Lung, Colorectal, Cervix, Endometrial, Ovarian, Prostate, Melanoma, Haematological. Other.

BMI (</≥ 23 kg/m²).

Hb (</≥ 12 g/dl).

Alb (20–35/>35 g/l).

Treatment plan modification.

BMI associated with outcome, p = 0.029 [no CI].

3.5/10

Risk of bias in data collection, selection bias, Data presentation.

Rajasekaran,

Singapore [70]

≥70 YO

Median 77

Range 70–94

n = 244

F = 95

M = 149

Gastrointestinal, Lung, Genitourinary, Other.

BMI (</≥27.5 kg/m²).

Hb (</≥ 12 g/dl).

Dominant handgrip (per kg increase).

Caregiver burden.

Hb associated with outcome

7.5/10

Risk of confounding, study design.

Randomised controlled trials	Age, years	Sample size, gender	Cancer diagnoses and treatment	Malnutrition proxy markers	Patient outcomes	Follow up	Study results [95% CI]	Quality score
Aparicio, France [71]	≥75 YO Mean 80 ± 3.7	n = 123 F = 57 M = 66	Metastatic colorectal cancer. Chemotherapy.	BMI (≤20/20–30/ ≥30 kg/m² Hb (</≥ 10 g/dl females, </≥ 11 g/dl males)	Dose intensity reduction ≥33%. Grade 3 to 4 toxicity. ≥1 Hospitalisation.	4 months after start of treatment.	No association between proxy markers and outcomes.	7.0/10 Risk of selection bias.
Falandry, France [72]	≥70 YO Median 79 Range 70–93	n = 98 F = 98 M = 0	Epithelial FIGO stage III or IV ovarian cancer. Chemotherapy.	Alb </≥ 35 g/l. BMI </≥21 kg/m²	Overall survival.	Median 17.4 months.	Alb associated with outcome in univariate analysis; HR 2.36, [no CI] p=0.003.	4.5/10 Risk of confounding, risk of bias in data collection.

Randomised controlled trials

Age, years

Sample size, gender

Cancer diagnoses and treatment

Malnutrition proxy markers

Patient outcomes

Follow up

Study results [95% CI]

Quality score

Aparicio,

France [71]

≥75 YO

Mean 80 ± 3.7

n = 123

F = 57

M = 66

Metastatic colorectal cancer.

Chemotherapy.

BMI (≤20/20–30/ ≥30 kg/m²

Hb (</≥ 10 g/dl females, </≥ 11 g/dl males)

Dose intensity reduction ≥33%.

Grade 3 to 4 toxicity.

≥1 Hospitalisation.

4 months after start of treatment.

No association between proxy markers and outcomes.

7.0/10

Risk of selection bias.

Falandry,

France [72]

≥70 YO

Median 79

Range 70–93

n = 98

F = 98

M = 0

Epithelial FIGO stage III or IV ovarian cancer.

Chemotherapy.

Alb </≥ 35 g/l.

BMI </≥21 kg/m²

Overall survival.

Median 17.4 months.

Alb associated with outcome in univariate analysis; HR 2.36, [no CI] p=0.003.

4.5/10

Risk of confounding, risk of bias in data collection.

n number, F female, M male, YO years old, OR odds ratios, HR hazard ratios, CI confidence intervals, NA not applicable Alb albumin, Hb haemoglobin, CRP C-reactive protein, BMI body mass index NRI nutrition risk index, PNI prognostic nutritional index, GNRI geriatric nutrition risk index, CONUT controlling nutritional status score ADL activities of daily living.